File: obo-syntax.html

package info (click to toggle)
owlapi 5.1.20-1
links: PTS, VCS
area: main
in suites: bookworm, forky, sid, trixie
size: 21,600 kB
sloc: java: 149,988; xml: 8,139; makefile: 2
file content (2970 lines) | stat: -rw-r--r-- 128,964 bytes
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
  <title>OBO Flat File Format 1.4 Syntax and Semantics [DRAFT]</title>
  <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
  <style type="text/css">
/* Style for a "Working Group Note" */

/*
   Copyright 1997-2003 W3C (MIT, ERCIM, Keio). All Rights Reserved.
   The following software licensing rules apply:
   http://www.w3.org/Consortium/Legal/copyright-software */

/* $Id: base.css,v 1.25 2006/04/18 08:42:53 bbos Exp $ */

body {
  padding: 2em 1em 2em 70px;
  margin: 0;
  font-family: sans-serif;
  color: black;
  background: white;
  background-position: top left;
  background-attachment: fixed;
  background-repeat: no-repeat;
}
:link { color: #00C; background: transparent }
:visited { color: #609; background: transparent }
a:active { color: #C00; background: transparent }

a:link img, a:visited img { border-style: none } /* no border on img links */

a img { color: white; }        /* trick to hide the border in Netscape 4 */
@media all {                   /* hide the next rule from Netscape 4 */
  a img { color: inherit; }    /* undo the color change above */
}

th, td { /* ns 4 */
  font-family: sans-serif;
}

h1, h2, h3, h4, h5, h6 { text-align: left }
/* background should be transparent, but WebTV has a bug */
h1, h2, h3 { color: #005A9C; background: white }
h1 { font: 170% sans-serif }
h2 { font: 140% sans-serif }
h3 { font: 120% sans-serif }
h4 { font: bold 100% sans-serif }
h5 { font: italic 100% sans-serif }
h6 { font: small-caps 100% sans-serif }

.hide { display: none }

div.head { margin-bottom: 1em }
div.head h1 { margin-top: 2em; clear: both }
div.head table { margin-left: 2em; margin-top: 2em }

p.copyright { font-size: small }
p.copyright small { font-size: small }

@media screen {  /* hide from IE3 */
a[href]:hover { background: #ffa }
}

pre { margin-left: 2em }
/*
p {
  margin-top: 0.6em;
  margin-bottom: 0.6em;
}
*/
dt, dd { margin-top: 0; margin-bottom: 0 } /* opera 3.50 */
dt { font-weight: bold }

pre, code { font-family: monospace } /* navigator 4 requires this */

ul.toc, ol.toc {
  list-style: disc;		/* Mac NS has problem with 'none' */
  list-style: none;
}

@media aural {  
  h1, h2, h3 { stress: 20; richness: 90 }
  .hide { speak: none }
  p.copyright { volume: x-soft; speech-rate: x-fast }
  dt { pause-before: 20% }
  pre { speak-punctuation: code } 
}



/* Copyright 1997-2003 W3C (MIT, ERCIM, Keio). All Rights Reserved.
   The following software licensing rules apply:
   http://www.w3.org/Consortium/Legal/copyright-software */

body {
//  background-image: url(http://www.w3.org/StyleSheets/TR/logo-WG-Note);
//background-image: url(http://www.obofoundry.org/images/foundrylogo.gif);
}


  </style>
  <link href="owl.css" rel="stylesheet" type="text/css" />

<body>

<div class="head">
<a class="image" style="float:right" href="/"
     title="The Open Biological and Biomedical Ontologies"><img class="middle"
						 src="http://www.obofoundry.org/images/foundrylogo.gif"
						 width="64"
						 alt="The Open Biological and Biomedical Ontology Foundry" /></a>
<h1 class="mainTitle">
OBO Flat File Format 1.4 Syntax and Semantics [WORKING DRAFT]</h1>
<h2><a id="editordraft" name="editordraft"></a>Editor's Draft: May 2012</h2>
<dl>
  <dt>This version: </dt>  <dd><a href="http://purl.obolibrary.org/obo/oboformat/2011-11-29/doc/obo-syntax.html">http://purl.obolibrary.org/obo/oboformat/2011-11-29/doc/obo-syntax.html</a></dd>
  <dt>Editor's version: </dt>  <dd><a href="http://oboformat.googlecode.com/svn/trunk/doc/obo-syntax.html">http://oboformat.googlecode.com/svn/trunk/doc/obo-syntax.html</a></dd>
  <dt>Latest version: </dt>  <dd><a href="http://purl.obolibrary.org/obo/oboformat/spec.html">http://purl.obolibrary.org/obo/oboformat/spec.html</a></dd>
  <dt>Previous version: </dt>  <dd><a href="http://purl.obolibrary.org/obo/oboformat/2011-08-28/doc/obo-syntax.html">http://purl.obolibrary.org/obo/oboformat/2011-08-28/doc/obo-syntax.html</a></dd>
 <dt>Editors: </dt>
  <dd>Chris Mungall, Lawrence Berkeley National Laboratory</dd>
  <dd>Alan Ruttenberg, University at Buffalo</dd>
  <dd>Ian Horrocks, University of Oxford</dd>
  <dd>David Osumi-Sutherland, Cambridge University</dd>
 <dt>Contributors: </dt>
  <dd>Erick Antezana, Bayer CropScience</dd>
  <dd>James Balhoff, National Evolutionary Synthesis Center</dd>
  <dd>Melanie Courtot,  British Columbia Cancer Research Center</dd>
  <dd>Heiko Dietze, Lawrence Berkeley National Laboratory</dd>
  <dd>John Day-Richter, Google</dd>
  <dd>Mathew Horridge, University of Manchester</dd>
  <dd>Amelia Ireland, The Jackson Laboratory</dd>
  <dd>Suzanna Lewis, Lawrence Berkeley National Laboratory</dd>
  <dd>Shahid Manzoor, Lawrence Berkeley National Laboratory</dd>
  <dd>Syed Hamid Tirmizi, University of Texas</dd>
</dl>
<hr>
</div>

<h2 class="notoc"><a id="abstract" name="abstract">Abstract</a></h2>
<div class="abstract">
<p>This document provides a BNF specification and mapping to OWL2 of
  the <a href="http://www.geneontology.org/GO.format.obo-1_4.shtml">OBO
    Flat File Format, version 1.4</a></p>
</div>

<div class="status">
<h2 class="notoc"><a id="status" name="status">Status of this Document</a></h2>
          <p><strong>This is an working draft, for comment by the community.</strong></p>
          <p>Comments should be sent to <a href="mailto:obo-format@lists.sourceforge.net">obo-format@lists.sourceforge.net</a> (<a href="http://sourceforge.net/mailarchive/forum.php?forum=obo-format">archive</a>)
            to ensure wide visibility. </p>
          <p>
            Most of this document can be regarded as stable. The parts that are likely to be unstable <span style="color:red">are marked red</span>. 
          </p>
          <p>
            There is a working reference implementation available from <a href="http://oboformat.googlecode.com/">http://oboformat.googlecode.com/</a>.
          </p>
</div>
<hr>



<div class="toc">
<h2 class="notoc"><a id="toc" name="toc">Table of Contents</a></h2>

<ul class="toc">
  <li class="tocline1"><a href="#1">1 Introduction</a></li>
  <li class="tocline1"><a href="#2">2 Preliminary Definitions</a>
    <ul class="toc">
      <li class="tocline2"><a href="#2.1">2.1 BNF Notation</a></li>
      <li class="tocline2"><a href="#2.2">2.2 Spacing Characters</a></li>
      <li class="tocline2"><a href="#2.3">2.3 Line Termination</a></li>
    </ul>
  </li>
  <li class="tocline1"><a href="#3">3 OBO Grammar</a></li>
  <li class="tocline1"><a href="#4">4 OBO Document Structure</a></li>
  <li class="tocline1"><a href="#5">5 OBO Semantics: Mapping to OWL2-DL</a></li>
  <li class="tocline1"><a href="#6">6 OBO Sublanguages</a></li>
  <li class="tocline1"><a href="#7">7 OWL Macros</a></li>
  <li class="tocline1"><a href="#8">8 Recommendations (informative)</a></li>
  <li class="tocline1"><a href="#9">9 References</a></li>
</ul>
<hr>
</div>



<h2><a name="1"></a>1 Introduction</h2>

<p>This document specifies the syntax and semantics of OBO Format (OBOF)</p>

<p>
  The first part of this specification describes the syntax and
  structure of OBO Format using BNF. This states how strings of
  characters in OBOF Files are parsed into abstract OBO documents, and
  describes constraints on the structure of these documents. The
  second part of this specification describes the semantics of an OBO
  document via a mapping to OWL2-DL and community metadata
  vocabularies.
</p>

<h2><a name="1.1"></a>1.1 Relationship to previous specifications and translations</h2>

<p>
This specification builds on previous attempts to specify the syntax
            and/or semantics of OBOF.
</p>

<h2><a name="1.1.1"></a>1.1.1 Golbreich and Horrocks translation</h2>
<p>
  This document is derived from and indebted to a previous document by
  Ian Horrocks [<cite><a href="#obo-owl-horrocks" title="">OBO-OWL
  Horrocks</a></cite>], later described in an accompanying paper
  [<cite><a href="#obo-owl-golbreich" title="">OBO-OWL
  Golbreich</a></cite>]. This document provided a semantics for OBO
  Format in terms of OWL-DL and what was then called OWL1.1 (later to
  become OWL2). The syntax was underspecified and the mapping was
  incomplete as there was no treatment of annotation properties.
</p>
<p>
<b>implementation:</b> This mapping was implemented in the OWLAPI, and
            is consequently used by current versions of Protege4
</p>

<h2><a name="1.1.2"></a>1.1.2 OboInOwl</h2>
<p>
  The OboInOwl mapping attempted to fill in the gaps in the Horrocks
  translation by providing a full translation of all of OBOF 1.2,
  including synonyms and definitions[<cite><a href="#obo-owl-ncbo"
  title="">OBO-OWL NCBO</a></cite>]. This mapping requires use of the
  n-ary relations pattern to represent OBO metadata in its entirety.
  This has undesirable side-effects, such as introducing modeling
  classes and individuals into an ontology, affecting both reasoning
  and usability.
</p>
<p>
<b>implementation:</b>This mapping was implemented via an XSLT
  translation from OBO-XML. All class and ontology URIs are of the
  form <b>http://purl.org/obo/owl</b>.
</p>

<h2><a name="1.1.3"></a>1.1.3. OBO Format 1.3 mapping</h2>
<p>
  OBO Format 1.3 was accompanied by a grammar and mapping to FOL
  [<cite><a href="#obolog" title="">Obolog</a></cite>]. This mapping
  was based on the 2005 version of the OBO Relation Ontology and
  OBOF1.3/Obolog attempted to do justice to dual instance/type level
  relations and ternary temporally-qualified instance-level
  relations. However, this introduced too large an impedance mismatch
  between OBOF and OWL, and as a consequence OBOF1.3 and Obolog have
  been <b>deprecated</b>.
</p>
<p>
<b>implementation:</b>Deprecated.
</p>

<h2><a name="1.1.4"></a>1.1.4 Tirmizi et al</h2>
<p>
  More recent work by Tirmizi provides an implementation of a fully
  roundtrippable mapping between OBO to OWL and back
  [<cite><a href="#obo-owl-swat4ls" title="">OBO-OWL
  SWAT4LS</a></cite>] [<cite><a href="#obo-owl-jbms" title="">OBO-OWL
  JBMS</a></cite>]. This did not attempt to provide a full grammar,
  and implemented the same specification as <a href="#1.1.2">1.1.2</a>
</p>
<p>
<b>implementation:</b>OBO-Edit 2.0 (removed from OBO-Edit-2.1)
</p>

<h2><a name="1.1.5"></a>1.1.5 OBO Format Guides</h2>
<p>
The OBO Format 1.2 guide is too loose to be regarded as a full
            specification. It served as a basis for
            (<a href="#1.1.1">1.1.1</a>).
</p>
<p>
The [<cite><a href="#obo-1-4" title="">OBO Format 1.4
                guide</a></cite>] accompanies this document. It is
                intended as a guide, not a normative specification.
</p>
<h2><a name="1.1.6"></a>1.1.6 Normative mapping</h2>
<p>
  This document supersedes previous efforts. The goal is to provide a
  complete, normative specification of both the syntax and semantics
  of OBOF1.4. The syntax and semantics specified here also hold for
  any OBOF1.2 document.
</p>
<p>
  The mapping provided here is consistent with the original
  Golbreich/Horrocks translation (<a href="#1.1.1">1.1.1</a>). It improves
  on this translation by specifying the grammar in full, and by
  providing a full mapping of all OBOF constructs to OWL2, through a
  combined use of the Information Artefact Ontology (IAO) and the
  OboInOwl vocabulary.
</p>
<p>
  This mapping uses a different means of translating OBOF
  terminological elements (synonyms, definitions, etc) to OWL2 than
  the one provided in <a href="#1.1.2">1.1.2</a>. The n-ary relations
  pattern has been abandoned, and replaced by the use of OWL2
  AxiomAnnotations. Many of the OboInOwl vocabulary elements have been
  replaced by IAO annotation properties. This mapping also uses a
  different URI scheme. In place of the
  deprecated <b>http://purl.org/obo/owl</b> URIs, the OBO Foundry
  standard <b>http://purl.obolibrary/org/obo</b> scheme is now used.
</p>
<p>
  The semantics specified in this document are inconsistent with the
  abandoned OBOF1.3 specification <a href="#1.1.3">1.1.3</a>, which is
  strongly deprecated.
</p>
<p>
  This specification also introduces new constructs absent from
  previous versionsof OBOF, and clarifies many syntactic and semantic
  elements that were previously unclear.
</p>
<p>
<b>implementation:</b> the reference implementation of this ontology
            can be found at <a href="http://oboformat.googlecode.com">the oboformat site</a>
</p>

<h2><a name="1.2"></a>1.2 Additional introductory material</h2>

<p>
More material can be found
at <a href="http://oboformat.googlecode.com">the oboformat site</a>
</p>

<a name="Preliminary_Definitions"></a><h2> <span class="mw-headline">2  Preliminary Definitions </span></h2>

<a name="BNF"></a><h2> <span class="mw-headline">2.1 BNF Notation </span></h2>

<p>OBO-Format 1.4 is defined using a standard BNF notation, which is summarized in the table below. 
</p>

<div class="center">
<table border="1">
<caption> <span class="caption">Table 1.</span> The BNF Notation Used in this Document
</caption><tr>
<th> Construct
</th><th> Syntax
</th><th> Example
</th></tr>
<tr>
<td> non-terminal symbols
</td><td> boldface
</td><td> <span class="nonterminal">QuotedString</span>
</td></tr>
<tr>
<td> terminal symbols
</td><td> single quoted
</td><td> <span class="name">'Term'</span>
</td></tr>
<tr>
<td> zero or more
</td><td> curly braces
</td><td> <span class="name">{ </span><span class="nonterminal">entity-frame</span><span class="name"> }</span>
</td></tr>
<tr>
<td> zero or one
</td><td> square brackets
</td><td> <span class="name">[ </span><span class="nonterminal">ws SynonymType-ID</span><span class="name"> ]</span>
</td></tr>
<tr>
<td> alternative
</td><td> vertical bar
</td><td> <span class="nonterminal">Class-ID</span><span class="name"> | </span><span class="nonterminal">Relation-ID</span>
</td></tr>
<tr>
<td> grouping
</td><td> parentheses
</td><td> ( <span class="nonterminal"></span> )
</td></tr>
<tr>
<td> complementation</td>
<td> minus symbol</td>
<td> ( <span class="nonterminal">character</span><span class="name"> - </span><span class="nonterminal">NewLineChar</span> )
</td></tr>
</table>
</div>

<p>Because generic tag-value pairs occur in very many places in the syntax,
to save space the grammar has meta-productions for basic Tag-Values
  Pairs (TVPs), Boolean Tags (BTs) and for the tags themselves (T):
</p>
<p class="grammar" style="white-space: pre;"><span class="nonterminal">&lt;T&gt;-TVP</span>  ::= '<span class="terminal">&lt;T&gt;:</span>' [<span class="nonterminal">ws</span>] <span class="nonterminal">UnquotedString</span>
<span class="nonterminal">&lt;T&gt;-BT</span>   ::= '<span class="terminal">&lt;T&gt;:</span>' [<span class="nonterminal">ws</span>] ( 'true' | 'false' )
<span class="nonterminal">&lt;T&gt;-Tag</span>  ::= '<span class="terminal">&lt;T&gt;:</span>' [<span class="nonterminal">ws</span>]
</p>

<p>We provide an additional List meta-production, for comma-delimited lists
</p>
<p class="grammar" style="white-space: pre;"><span class="nonterminal">&lt;NT&gt;List</span>  ::= [<span class="nonterminal">&lt;NT&gt;</span> {',' <span class="nonterminal">ws</span> <span class="nonterminal">&lt;NT&gt;</span> }]
</p>
<p>
For example, <span class="nonterminal">QualifierList</span> should be
interpreted as: [ <span class="nonterminal">Qualifier</span> { ','
<span class="nonterminal">ws</span> <span class="nonterminal">Qualifier</span>} ]
</p>


<p>
Documents in the obo-format consist of sequences of Unicode characters
[<cite><a href="#ref-unicode" title="">UNICODE</a></cite>] and are encoded in UTF-8
[<cite><a href="#ref-rfc-3629" title="">RFC 3629</a></cite>].
</p>



<h3><a name="2.2"></a>2.2 Characters</h3>

<h4><a name="2.2.0"></a>2.2.0 Basic Characters</h4>

<p class="grammar" style="white-space: pre;">

<span class="nonterminal">Alpha-Char</span> ::=  <span class="nonterminal">a</span> |<span class="nonterminal">b</span> |<span class="nonterminal">c</span> |<span class="nonterminal">d</span> |<span class="nonterminal">e</span> |<span class="nonterminal">f</span> |<span class="nonterminal">g</span> |<span class="nonterminal">h</span> |<span class="nonterminal">i</span> |<span class="nonterminal">j</span> |<span class="nonterminal">k</span> |<span class="nonterminal">l</span> |<span class="nonterminal">m</span> |<span class="nonterminal">n</span> |<span class="nonterminal">o</span> |<span class="nonterminal">p</span> |<span class="nonterminal">q</span> |<span class="nonterminal">r</span> |<span class="nonterminal">s</span> |<span class="nonterminal">t</span> |<span class="nonterminal">u</span> |<span class="nonterminal">v</span> |<span class="nonterminal">w</span> |<span class="nonterminal">x</span> |<span class="nonterminal">y</span> |<span class="nonterminal">z</span> | <span class="nonterminal">A</span> |<span class="nonterminal">B</span> |<span class="nonterminal">C</span> |<span class="nonterminal">D</span> |<span class="nonterminal">E</span> |<span class="nonterminal">F</span> |<span class="nonterminal">G</span> |<span class="nonterminal">H</span> |<span class="nonterminal">I</span> |<span class="nonterminal">J</span> |<span class="nonterminal">K</span> |<span class="nonterminal">L</span> |<span class="nonterminal">M</span> |<span class="nonterminal">N</span> |<span class="nonterminal">O</span> |<span class="nonterminal">P</span> |<span class="nonterminal">Q</span> |<span class="nonterminal">R</span> |<span class="nonterminal">S</span> |<span class="nonterminal">T</span> |<span class="nonterminal">U</span> |<span class="nonterminal">V</span> |<span class="nonterminal">W</span> |<span class="nonterminal">X</span> |<span class="nonterminal">Y</span> |<span class="nonterminal">Z</span>
<span class="nonterminal">Digit</span> ::= <span class="nonterminal">0</span> |<span class="nonterminal">1</span> |<span class="nonterminal">2</span> |<span class="nonterminal">3</span> |<span class="nonterminal">4</span> |<span class="nonterminal">5</span> |<span class="nonterminal">6</span> |<span class="nonterminal">7</span> |<span class="nonterminal">8</span> |<span class="nonterminal">9</span> 

</p>


<h4><a name="2.2.1"></a>2.2.1 Spacing Characters</h4>

<p class="grammar" style="white-space: pre;">
<span class="nonterminal">WhiteSpaceChar</span> ::= ' ' | \t | U+0020 | U+0009
<span class="nonterminal">ws</span> ::= <span class="nonterminal">WhiteSpaceChar</span> { <span class="nonterminal">WhiteSpaceChar</span> }
<span class="nonterminal">NewlineChar</span> ::= \r | \n | U+000A | U+000C | U+000D
<span class="nonterminal">nl</span> ::= [<span class="nonterminal">ws</span>] <span class="nonterminal">NewLineChar</span>
<span class="nonterminal">nl*</span> ::= { <span class="nonterminal">nl</span> }
</p>

<h4><a name="2.2.2"></a>2.2.2 Special Characters</h4>

<p>
We use the nonterminal <span class="nonterminal">UniCodeChar</span> to
represent any Unicode character.
</p>

<p>
An <span class="nonterminal">OBOChar</span> is either any non-newline
unicode character, or a unicode character escaped by a backslash. This
can be used to encode newline characters and other special characters
in different contexts.
</p>

<p class="grammar" style="white-space: pre;">
<span class="nonterminal">UniCodeChar</span> ::= <i>any Unicode character</i>
<span class="nonterminal">OBOChar</span> ::= '\' <span class="nonterminal">Alpha-Char</span> | ( <span class="nonterminal">UniCodeChar</span> - (<span class="nonterminal">NewLineChar</span> | '\') )
<span class="nonterminal">NonWsChar</span> ::=  ( <span class="nonterminal">OBOChar</span> - <span class="nonterminal">WhiteSpaceChar</span> ) 
</p>


<h3><a name="2.3"></a>2.3 Line Termination</h3>

<p>
Each tag-value clause in OBOF is line separated. The line can
optionally be ended by a HiddenComment, indicated by the '!' character
- this is semantically silent can be ignored by the parser. Each
clause can also have zero or more comma-separated
tag-value <i>trailing qualifiers</i> between a '{' and a '}' - this is
called a QualifierBlock.
</p>

<p class="grammar" style="white-space: pre;">
<span class="nonterminal">EOL</span> ::=  {<span class="nonterminal">WhiteSpaceChar</span>} [ <span class="nonterminal">QualifierBlock</span> ] {<span class="nonterminal">WhiteSpaceChar</span>} [ <span class="nonterminal">HiddenComment</span> ] {<span class="nonterminal">WhiteSpaceChar</span>} <span class="nonterminal">NewLineChar</span>
<span class="nonterminal">HiddenComment</span> ::= '!' { ( <span class="nonterminal">UniCodeChar</span> - <span class="nonterminal">NewlineChar</span> ) } 
<span class="nonterminal">QualifierBlock</span> ::= '{' <span class="nonterminal">QualifierList</span> '}' 
<span class="nonterminal">QualifierList</span>  ::= <span class="nonterminal">Qualifier</span> {',' <span class="nonterminal">ws</span> <span class="nonterminal">Qualifier</span> }
<span class="nonterminal">Qualifier</span> ::= <span class="nonterminal">Rel-ID</span> '=' <span class="nonterminal">QuotedString</span> 
</p>

<h3><a name="2.4"></a>2.4 Clause Values</h3>

<p>
Depending on the tag, the values for clauses may be written as
UnquotedStrings (in which case the entire line after the tag and
before any qualifier block is used as the clause value), or as a
QuotedString, in which case it is enclosed within double quotes.
</p>

<p class="grammar" style="white-space: pre;">
<span class="nonterminal">QuotedString</span> ::=  <span class="nonterminal">DblQuote</span> { ( <span class="nonterminal">OBOChar</span> - <span class="nonterminal">DblQuote</span> ) } <span class="nonterminal">DblQuote</span>
<span class="nonterminal">UnquotedString</span> ::=  { <span class="nonterminal">OBOChar</span>  }
</p>

<h3><a name="2.5"></a>2.5 Identifiers</h3>

<p>
We introduce 3 rules for class IDs, relation IDs and instance IDs -
these all have the same structure.
</p>

<p>
IDs fall into one of 3 categories: unprefixed, prefixed or URLs. The
prefixed category is divided into two: canonical and non-canonical
prefixed IDs. Whilst canonical prefixed IDs are preferred, many OBO
documents make use of non-canonical prefixed IDs or unprefixed IDs in
a variety of contexts - relation, subsetdef and synonymtypedef IDs
have traditionally been unprefixed; xrefs frequently do not conform to
the canonical prefixed ID pattern.
</p>

<p class="grammar" style="white-space: pre;">
<span class="nonterminal">Class-ID</span> ::=  <span class="nonterminal">ID</span>
<span class="nonterminal">Rel-ID</span> ::=  <span class="nonterminal">ID</span>
<span class="nonterminal">Instance-ID</span> ::=  <span class="nonterminal">ID</span>
<!-- TODO: URLs -->
<span class="nonterminal">ID</span> ::=  <span class="nonterminal">Prefixed-ID</span> | <span class="nonterminal">Unprefixed-ID</span> | <span class="nonterminal">URL-as-ID</span>

<span class="nonterminal">URL-as-ID</span> ::=   ( <span class="terminal">'http:'</span> | <span class="terminal">'https:'</span> ) { <span class="nonterminal">NonWsChar</span> }
<span class="nonterminal">Unprefixed-ID</span> ::=  { ( <span class="nonterminal">NonWsChar</span> - <span class="terminal">':'</span> ) }
<span class="nonterminal">Prefixed-ID</span> ::=  <span class="nonterminal">Canonical-Prefixed-ID</span> | <span class="nonterminal">NonCanonical-Prefixed-ID</span> 
</p>

<p class="grammar" style="white-space: pre;">

<span class="nonterminal">Canonical-IDPrefix</span> ::=  <span class="nonterminal">Alpha-Char</span> { ( '_' | <span class="nonterminal">Alpha-Char</span> ) }
<span class="nonterminal">Canonical-LocalID</span> ::=  { <span class="nonterminal">Digit</span> }
<span class="nonterminal">Canonical-Prefixed-ID</span> ::=  <span class="nonterminal">Canonical-IDPrefix</span> ':' <span class="nonterminal">Canonical-LocalID</span> 
<span class="nonterminal">NonCanonical-Prefixed-ID</span> ::=  ( <span class="nonterminal">Any-IDPrefix</span> ':' <span class="nonterminal">Any-LocalID</span> ) - <span class="nonterminal">Canonical-Prefixed-ID</span>
<span class="nonterminal">Any-IDPrefix</span> ::=  { ( <span class="nonterminal">NonWsChar</span> - <span class="nonterminal">':'</span> ) }
<span class="nonterminal">Any-LocalID</span> ::=  { <span class="nonterminal">NonWsChar</span> }

</p>


<h3><a name="2.6"></a>2.6 Xref Lists</h3>

            <p>
              An XrefList is a comma-separated list of zero or more
              Xrefs inside square brackets (these are manditory). Each
              Xref is a string of characters, optionally followed by a
              auoted description. Note that commas in the xref need
              to be escaped. We include an extra production rule for
              when solitary Xrefs are provided (i.e. with the 'xref'
              tag) - in this case commas do not need to be escaped.
            </p>

<p class="grammar" style="white-space: pre;">

<span class="nonterminal">XrefList</span> ::=  '[' [ <span class="nonterminal">Xref</span> { ',' <span class="nonterminal">ws</span> <span class="nonterminal">XrefNoComma</span> } ] { <span class="nonterminal">ws</span> } ']'
<span class="nonterminal">XrefNoComma</span> ::= <span class="nonterminal">XrefChar</span> {  <span class="nonterminal">XrefChar</span> }  [ <span class="nonterminal">ws</span> <span class="nonterminal">QuotedString</span> ]
<span class="nonterminal">XrefChar</span> ::= (<span class="nonterminal">NonWsChar</span> - ',')
<span class="nonterminal">Xref</span> ::= <span class="nonterminal">ID</span>  [ <span class="nonterminal">ws</span> <span class="nonterminal">QuotedString</span> ]

</p>

<h2><a name="3"></a>3 OBO Grammar</h2>
<h3><a name="3.1"></a>3.1 OBO Document Structure</h3>

<p>An OBO document consists of a series of header frame followed by
zero or more entity frames. Each frame has a set of clauses, or
&lt;tag,value&gt; pairs. A tag is a token which may be drawn from the
set of defined OBOF tags. The value can be atomic or multi-values. In
the OBO Abstract syntax a clause is written
&lt;Tag&gt;(&lt;V1&gt;...&lt;Vn&gt;).</p>

<p>Each frame should by convention be separated by an empty line, but this is not enforced in the syntax</p>

<p>Note that the term "frame" replaces the previously used "stanza" in
order to be more consistent with Manchester Syntax.</p>

<p class="grammar" style="white-space: pre;">
  <span class="nonterminal">OBO-Doc</span> := <span class="nonterminal">header-frame</span> { <span class="nonterminal">entity-frame</span> } <span class="nonterminal">nl*</span>
</p>


<h3><a name="3.2"></a>3.2 OBO Headers</h3>



<p class="grammar" style="white-space: pre;">
<span class="nonterminal">header-frame</span> ::= { <span class="nonterminal">header-clause</span> <span class="nonterminal">nl*</span> }
<span class="nonterminal">header-clause</span> ::= <span class="nonterminal">format-version-TVP</span>
        | <span class="nonterminal">data-version-TVP</span>
        | <span class="nonterminal">date-Tag</span>  <span class="nonterminal">DD:MM:YYYY</span> <span class="nonterminal">sp</span> <span class="nonterminal">hh:mm</span>
        | <span class="nonterminal">saved-by-TVP</span>
        | <span class="nonterminal">auto-generated-by-TVP</span>
        | <span class="nonterminal">import-Tag</span>  <span class="nonterminal">IRI</span> | <span class="nonterminal">filepath</span>
        | <span class="nonterminal">subsetdef-Tag</span>  <span class="nonterminal">ID</span> <span class="nonterminal">sp</span> <span class="nonterminal">QuotedString</span>
        | <span class="nonterminal">synonymtypedef-Tag</span>  <span class="nonterminal">ID</span> <span class="nonterminal">sp</span> <span class="nonterminal">QuotedString</span>
             [  <span class="nonterminal">SynonymScope</span> ]
        | <span class="nonterminal">default-namespace-Tag</span>  <span class="nonterminal">OBONamespace</span>
        | <span class="nonterminal">idspace-Tag</span> <span class="nonterminal">IDPrefix</span> <span class="nonterminal">sp</span> <span class="nonterminal">IRI</span>
            [ <span class="nonterminal">sp</span> <span class="nonterminal">QuotedString</span> ]
        | <span class="nonterminal">treat-xrefs-as-equivalent-Tag</span> <span class="nonterminal">IDPrefix</span>
        | <span class="nonterminal">treat-xrefs-as-genus-differentia-Tag</span> <span class="nonterminal">IDPrefix</span> <span class="nonterminal">ws</span> <span class="nonterminal">Rel-ID</span> <span class="nonterminal">ws</span> <span class="nonterminal">Class-ID</span>
        | <span class="nonterminal">treat-xrefs-as-reverse-genus-differentia-Tag</span> <span class="nonterminal">IDPrefix</span> <span class="nonterminal">ws</span> <span class="nonterminal">Rel-ID</span> <span class="nonterminal">ws</span> <span class="nonterminal">Class-ID</span>
        | <span class="nonterminal">treat-xrefs-as-relationship-Tag</span> <span class="nonterminal">IDPrefix</span> <span class="nonterminal">Rel-ID</span>
        | <span class="nonterminal">treat-xrefs-as-is_a-Tag</span> <span class="nonterminal">IDPrefix</span>
        | <span class="nonterminal">treat-xrefs-as-has-subclass-Tag</span> <span class="nonterminal">IDPrefix</span>
        | <span style="color:red"><span class="nonterminal">property_value-Tag</span> <span class="nonterminal">Relation-ID</span> ( <span class="nonterminal">QuotedString</span> <span class="nonterminal">XSD-Type</span> | <span class="nonterminal">ID</span> ) {<span class="nonterminal">WhiteSpaceChar</span>} [ <span class="nonterminal">QualifierBlock</span> ] {<span class="nonterminal">WhiteSpaceChar</span>} [ <span class="nonterminal">HiddenComment</span> ]</span>
        | <span class="nonterminal">remark-TVP</span>
        | <span class="nonterminal">ontology-TVP</span>
        | <span class="nonterminal">owl-axioms-TVP</span>
        | <span class="nonterminal">UnreservedToken</span> ':'  [  <span class="nonterminal">ws</span> ] <span class="nonterminal">UnquotedString</span>
<span class="nonterminal">entity-frame</span> ::= <span class="nonterminal">term-frame</span> | <span class="nonterminal">typedef-frame</span> | <span class="nonterminal">instance-frame</span> 
</p>


<h3><a name="3.3"></a>3.3 Term Frames</h3>

<p >Term frames introduce and define the meaning of <i>terms</i> (AKA
classes).</p >

<p class="grammar" style="white-space: pre;">
<span class="nonterminal">term-frame</span> ::= <span class="nonterminal">nl*</span>
        '[Term]' <span class="nonterminal">nl</span>
        <span class="nonterminal">id-Tag</span> <span class="nonterminal">Class-ID</span> <span class="nonterminal">EOL</span>
        { <span class="nonterminal">term-frame-clause</span> <span class="nonterminal">EOL</span> }

<span class="nonterminal">term-frame-clause</span> ::= 
        <span class="nonterminal">is_anonymous-BT</span>
        | <span class="nonterminal">name-TVP</span>
        | <span class="nonterminal">namespace-Tag</span> <span class="nonterminal">OBONamespace</span> 
        | <span class="nonterminal">alt_id-Tag</span> <span class="nonterminal">ID</span> 
        | <span class="nonterminal">def-Tag</span> <span class="nonterminal">QuotedString</span> <span class="nonterminal">ws</span> <span class="nonterminal">XrefList</span> 
        | <span class="nonterminal">comment-TVP</span>
        | <span class="nonterminal">subset-Tag</span> <span class="nonterminal">Subset-ID</span> 
        | <span class="nonterminal">synonym-Tag</span> <span class="nonterminal">QuotedString</span> <span class="nonterminal">ws</span> <span class="nonterminal">SynonymScope</span> [ <span class="nonterminal">ws</span> <span class="nonterminal">SynonymType-ID</span> ] <span class="nonterminal">XrefList</span> 
        | <span class="nonterminal">xref-Tag</span> <span class="nonterminal">Xref</span> 
        | <span class="nonterminal">builtin-BT</span>
        | <span class="nonterminal">property_value-Tag</span> <span class="nonterminal">Relation-ID</span> ( <span class="nonterminal">QuotedString</span> <span class="nonterminal">XSD-Type</span> | <span class="nonterminal">ID</span> )
        | <span class="nonterminal">is_a-Tag</span> <span class="nonterminal">Class-ID</span> 
        | <span class="nonterminal">intersection_of-Tag</span> <span class="nonterminal">Class-ID</span> 
        | <span class="nonterminal">intersection_of-Tag</span> <span class="nonterminal">Relation-ID</span> <span class="nonterminal">Class-ID</span> 
        | <span class="nonterminal">union_of-Tag</span> <span class="nonterminal">Class-ID</span> 
        | <span class="nonterminal">equivalent_to-Tag</span> <span class="nonterminal">Class-ID</span> 
        | <span class="nonterminal">disjoint_from-Tag</span> <span class="nonterminal">Class-ID</span> 
        | <span class="nonterminal">relationship-Tag</span> <span class="nonterminal">Relation-ID</span> <span class="nonterminal">Class-ID</span> 
        | <span class="nonterminal">is_obsolete-BT</span>
        | <span class="nonterminal">replaced_by-Tag</span> <span class="nonterminal">Class-ID</span> 
        | <span class="nonterminal">consider-Tag</span> <span class="nonterminal">ID</span> 
        | <span class="nonterminal">created_by-Tag</span> <span class="nonterminal">Person-ID</span> 
        | <span class="nonterminal">creation_date-Tag</span> <span class="nonterminal">ISO-8601-DateTime</span> 
</p >


<h3><a name="3.4"></a>3.4 Typedef Frames</h3>

<p >Typedef frames introduce and define the meaning of
<i>relations</i> (AKA properties).</p >

<p class="grammar" style="white-space: pre;">
<span class="nonterminal">typedef-frame</span> ::= [ <span class="nonterminal">nl</span> ]
        '[Typedef]' <span class="nonterminal">nl</span>
        <span class="nonterminal">id-Tag</span> <span class="nonterminal">Relation-ID</span> <span class="nonterminal">EOL</span>
        { <span class="nonterminal">typedef-frame-clause</span> <span class="nonterminal">EOL</span> }

<span class="nonterminal">typedef-frame-clause</span> ::= 
        <span class="nonterminal">is_anonymous-BT</span>
        | <span class="nonterminal">name-TVP</span>
        | <span class="nonterminal">namespace-Tag</span> <span class="nonterminal">OBONamespace</span> 
        | <span class="nonterminal">alt_id-Tag</span> <span class="nonterminal">ID</span> 
        | <span class="nonterminal">def-Tag</span> <span class="nonterminal">QuotedString</span> <span class="nonterminal">ws</span> <span class="nonterminal">XrefList</span> 
        | <span class="nonterminal">comment-TVP</span>
        | <span class="nonterminal">subset-Tag</span> <span class="nonterminal">Subset-ID</span> 
        | <span class="nonterminal">synonym-Tag</span> <span class="nonterminal">QuotedString</span> <span class="nonterminal">ws</span> <span class="nonterminal">SynonymScope</span> [ <span class="nonterminal">ws</span> <span class="nonterminal">SynonymType-ID</span> ] <span class="nonterminal">XrefList</span> 
        | <span class="nonterminal">xref-Tag</span> <span class="nonterminal">Xref</span> 
        | <span class="nonterminal">property_value-Tag</span> <span class="nonterminal">Relation-ID</span> ( <span class="nonterminal">QuotedString</span> <span class="nonterminal">XSD-Type</span> | <span class="nonterminal">ID</span> )
        | <span class="nonterminal">domain-Tag</span> <span class="nonterminal">Class-ID</span> 
        | <span class="nonterminal">range-Tag</span> <span class="nonterminal">Class-ID</span> 
        | <span class="nonterminal">builtin-BT</span>
        | <span class="nonterminal">holds_over_chain-Tag</span> <span class="nonterminal">Relation-ID</span> <span class="nonterminal">Relation-ID</span>
        | <span class="nonterminal">is_anti_symmetric-BT</span>
        | <span class="nonterminal">is_cyclic-BT</span>
        | <span class="nonterminal">is_reflexive-BT</span>
        | <span class="nonterminal">is_symmetric-BT</span>
        | <span class="nonterminal">is_transitive-BT</span>
        | <span class="nonterminal">is_functional-BT</span>
        | <span class="nonterminal">is_inverse_functional-BT</span>
        | <span class="nonterminal">is_a-Tag</span> <span class="nonterminal">Rel-ID</span> 
        | <span class="nonterminal">intersection_of-Tag</span> <span class="nonterminal">Rel-ID</span> 
        | <span class="nonterminal">union_of-Tag</span> <span class="nonterminal">Rel-ID</span> 
        | <span class="nonterminal">equivalent_to-Tag</span> <span class="nonterminal">Rel-ID</span> 
        | <span class="nonterminal">disjoint_from-Tag</span> <span class="nonterminal">Rel-ID</span> 
        | <span class="nonterminal">inverse_of-Tag</span> <span class="nonterminal">Rel-ID</span> 
        | <span class="nonterminal">transitive_over-Tag</span> <span class="nonterminal">Relation-ID</span>
        | <span class="nonterminal">equivalent_to_chain-Tag</span> <span class="nonterminal">Relation-ID</span> <span class="nonterminal">Relation-ID</span>
        | <span class="nonterminal">disjoint_over-Tag</span> <span class="nonterminal">Rel-ID</span> 
        | <span class="nonterminal">relationship-Tag</span> <span class="nonterminal">Rel-ID</span> <span class="nonterminal">Rel-ID</span>
        | <span class="nonterminal">is-obsolete-BT</span>   
        | <span class="nonterminal">replaced_by-Tag</span> <span class="nonterminal">Rel-ID</span> 
        | <span class="nonterminal">consider-Tag</span> <span class="nonterminal">ID</span> 
        | <span class="nonterminal">created_by-Tag</span> <span class="nonterminal">Person-ID</span> 
        | <span class="nonterminal">creation_date-Tag</span> <span class="nonterminal">ISO-8601-DateTime</span> 
        | <span class="nonterminal">expand_assertion_to-Tag</span> <span class="nonterminal">QuotedString</span> <span class="nonterminal">ws</span> <span class="nonterminal">XrefList</span> 
        | <span class="nonterminal">expand_expression_to-Tag</span> <span class="nonterminal">QuotedString</span> <span class="nonterminal">ws</span> <span class="nonterminal">XrefList</span> 
        | <span class="nonterminal">is_metadata_tag-BT</span>
        | <span class="nonterminal">is_class_level_tag-BT</span>
</p >


<h3><a name="3.5"></a>3.5 Instance Frames</h3>

<p >Instance frames introduce and define the meaning of
<i>instances</i> (AKA individuals).</p >

<p class="grammar" style="white-space: pre;">
<span class="nonterminal">instance-frame</span> ::= [ <span class="nonterminal">nl</span> ]
        '[Instance]' <span class="nonterminal">nl</span>
        <span class="nonterminal">id-Tag</span> <span class="nonterminal">Relation-ID</span> <span class="nonterminal">EOL</span>
        { <span class="nonterminal">instance-frame-clause</span> <span class="nonterminal">EOL</span> }

<span class="nonterminal">instance-frame-clause</span> ::= 
        <span class="nonterminal">is_anonymous-BT</span>
        | <span class="nonterminal">name-TVP</span>
        | <span class="nonterminal">namespace-Tag</span> <span class="nonterminal">OBONamespace</span> 
        | <span class="nonterminal">alt_id-Tag</span> <span class="nonterminal">ID</span> 
        | <span class="nonterminal">def-Tag</span> <span class="nonterminal">QuotedString</span> <span class="nonterminal">ws</span> <span class="nonterminal">XrefList</span> 
        | <span class="nonterminal">comment-TVP</span>
        | <span class="nonterminal">subset-Tag</span> <span class="nonterminal">Subset-ID</span> 
        | <span class="nonterminal">synonym-Tag</span> <span class="nonterminal">QuotedString</span> <span class="nonterminal">ws</span> <span class="nonterminal">SynonymScope</span> [ <span class="nonterminal">ws</span> <span class="nonterminal">SynonymType-ID</span> ] <span class="nonterminal">XrefList</span> 
        | <span class="nonterminal">xref-Tag</span> <span class="nonterminal">ID</span> 
        | <span class="nonterminal" style="color:red">property_value-Tag</span> <span class="nonterminal" style="color:red">Relation-ID</span> <span class="nonterminal" style="color:red">ID</span> 
        | <span class="nonterminal">instance_of-Tag</span> <span class="nonterminal">Class-ID</span> 
        | <span class="nonterminal" style="color:red">PropertyValueTagValue</span> 
        | <span class="nonterminal">relationship-Tag</span> <span class="nonterminal">Rel-ID</span> <span class="nonterminal">ID</span> 
        | <span class="nonterminal">created_by-Tag</span> <span class="nonterminal">Person-ID</span> 
        | <span class="nonterminal">creation_date-Tag</span> <span class="nonterminal">ISO-8601-DateTime</span> 
        | <span class="nonterminal">is-obsolete-BT</span>
        | <span class="nonterminal">replaced_by-Tag</span> <span class="nonterminal">Rel-ID</span> 
        | <span class="nonterminal">consider-Tag</span> <span class="nonterminal">ID</span> 
</p >


<h3><a name="3.5"></a>3.5 Synonym Scopes</h3>

<p class="grammar" style="white-space: pre;">
<span class="nonterminal">SynonymScope</span> ::= 
        'EXACT' | 'BROAD' | 'NARROW' | 'RELATED'
</p>


<h2><a name="4"></a>4 OBO Document Structure</h2>

<p>This section defines structural constraints on an OBO
Document. These structural constraints hold on an abstract OBO
Document - the result of parsing a physical OBO document file.</p>

<h3><a name="4.1"></a>4.1 Frame Identifiers</h3>

<p>
  Each frame in an abstract OBO document can be accessed by its
  identifier - the value of the <b>id</b> tag. A frame can only of one
  type. A document containing two frames of different types with
  the same id is illegal.
</p>

<h3><a name="4.1.1"></a>4.1.1 Merging frames during parsing</h3>

<p>
  Although it is strongly recommended that an OBOF document does not
  contain two frames with the same identifier, this is syntactically
  valid. If two frames have the same identifier, then they
  are <i>combined</i>. Only frames of the same type can be combined -
  if a document uses the same ID for two frames of different types,
  the document is structurally invalid.
</p>

<p>
  When two frames F1 and F2 are combined, the new frame F3 has a set
  of tag-values consisting of the union of the set of tag-values from
  F1 and the set of tag-values from F2. If two tag-values are have
  identical tags and identical values, they are considered a single
  tag-value.
</p>

<h3><a name="4.2"></a>4.2 Typedef constraints</h3>

              <p>
                If a Typedef frame has a clause is_metadata_tag(true),
                then that Typedef id may never be used as a Rel-ID in
                an intersection_of clause.
              </p>


<h3><a name="4.3"></a>4.3 OBO namespaces and ontology name</h3>

<p>
  Note that OBO namespaces are not the same as OWL namespaces - the
  analog of OWL namespaces are OBO ID spaces. OBO namespaces are
  semantics-free properties of a frame that allow partitioning of an
  ontology into sub-ontologies. For example, the GO is partitioned
  into 3 ontologies (3 OBO namespaces, 1 OWL namespace).
</p>

<p>
  Every frame must have exactly one namespace. However, these do not
  need to be explicitly assigned. After parsing an OBO Document, any
  frame without a namespace is assigned the default-namespace, from
  the OBO Document header. If this is not specified, the Parser
  assigns a namespace arbitrarily. It is recommended this is
  equivalent to the URL or file path from which the document was
  retrieved.
</p>

<p>
  Every OBODoc should have an "ontology" tag specified in the
  header. If this is not specified, then the parser should supply a
  default value. This value should be derived from the URL of the
  source of the ontology (typically using http or file schemes).
</p>

<h3><a name="4.4"></a>4.4 Processing Header Macros</h3>

<h4><a name="4.4.1"></a>4.4.1 Implicit Header Macros</h4>

<div>
  Every OBO-Doc is automatically populated with two header clauses:
<ul>
                  <li>treat-xrefs-as-equivalent(RO)</li>
                  <li>treat-xrefs-as-equivalent(BFO)</li>
</ul>
                It is illegal to override these. These declarations
                are taken as implicit, and an OBO format generator
                need not export these
</div>

<h4><a name="4.4.2"></a>4.4.2 Header Macro Translation</h4>

<p>
The following table shows how header macros are used to insert new
clauses into the ontology document. If the pre-condition is satisfied
(middle column) and the header contains the specified macro (first
column), then clauses are added to the ontology such that the
post-condition is satisfied (right hand column). Note that the
original xref tags are <i>not</i> removed from the source ontology.
</p>

<p>
These clauses are expanded into a fresh ontology, which can be
optionally imported.
</p>
<p>
In addition, any Typedef frames for relations used in a header macro
are also copied into the corresponding bridge ontology
</p>


<table border="1" cellspacing="0">
<caption><a id="header-macros" name="header-macros" ></a>
	Header Macros</caption>

<thead>
<tr><th>Header</th>
    <th>Precondition</th>
    <th>Add clauses</th>
</tr>
</thead>

<tbody>

  <tr>
    <td valign="top" style="white-space: pre;">treat-xrefs-as-equivalent(Prefix)</td>
    <td valign="top" style="white-space: pre;">xref(Prefix:B) &isin; Frame[A]</td>
    <td valign="top" style="white-space: pre;">equivalent_to(Prefix:B) &isin; Frame[A]</td>
  </tr>

  <tr>
    <td valign="top" style="white-space: pre;">treat-xrefs-as-is_a(Prefix)</td>
    <td valign="top" style="white-space: pre;">xref(Prefix:B) &isin; Frame[A]</td>
    <td valign="top" style="white-space: pre;">is_a(Prefix:B) &isin; Frame[A]</td>
  </tr>

  <tr>
    <td valign="top" style="white-space: pre;">treat-xrefs-as-has-subclass(Prefix)</td>
    <td valign="top" style="white-space: pre;">xref(Prefix:B) &isin; Frame[A]</td>
    <td valign="top" style="white-space: pre;">is_a(Prefix:A) &isin; Frame[B]</td>
  </tr>

  <tr>
    <td valign="top" style="white-space: pre;">treat-xrefs-as-genus-differentia(Prefix Rel-ID Class-ID)</td>
    <td valign="top" style="white-space: pre;">xref(Prefix:B) &isin; Frame[A] and
intersection_of(...) &notin; Frame[A]</td>
    <td valign="top" style="white-space: pre;">intersection_of(X:B) &isin; Frame[A] and
intersection_of(Rel-ID Class-ID) &isin; Frame[A]</td>
  </tr>

  <tr>
    <td valign="top" style="white-space: pre;">treat-xrefs-as-reverse-genus-differentia(Prefix Rel-ID Class-ID)</td>
    <td valign="top" style="white-space: pre;">xref(Prefix:B) &isin; Frame[A] and
intersection_of(...) &notin; Frame[B]</td>
    <td valign="top" style="white-space: pre;">intersection_of(X:A) &isin; Frame[B] and
intersection_of(Rel-ID Class-ID) &isin; Frame[B]</td>
  </tr>

  <tr>
    <td valign="top" style="white-space: pre;">treat-xrefs-as-relationship(Prefix Rel-ID)</td>
    <td valign="top" style="white-space: pre;">xref(Prefix:B) &isin; Frame[A]</td>
    <td valign="top" style="white-space: pre;">relationship(Rel-ID X:B) &isin; Frame[A]</td>
  </tr>


</tbody>
</table>

<h3><a name="4.4"></a>4.4 Tag Cardinality Constraints</h3>

<p>Note that the id tag is not specified in this table. Each frame has
  exactly one id tag, and the id tag uniquely identifier the
  frame. These constrainst must hold <i>after</i> frames with
  identical ids have been combined (see above).</p>

<p>Any tag not mentioned has free cardinality (zero, one or many).</p>

<table border="1" cellspacing="0">
<caption><a id="cardinality" name="cardinality" ></a>
	Cardinality Constraints</caption>

<thead>
<tr><th>Tag</th>
    <th>Frame</th>
    <th>Cardinality (Normative)</th>
    <th>Informative Notes</th>
</tr>
</thead>

<tbody>

  <tr><td>ontology</td><td>Header</td><td>zero or one</td><td>recommended for all docs</td></tr>
  <tr><td>format_version</td><td>Header</td><td>zero or one</td><td>-</td></tr>
  <tr><td>date</td><td>Header</td><td>zero or one</td><td>-</td></tr>
  <tr><td>default-namespace</td><td>Header</td><td>zero or one</td><td>-</td></tr>
  <tr><td>saved-by</td><td>Header</td><td>zero or one</td><td>-</td></tr>
  <tr><td>auto-generated-by</td><td>Header</td><td>zero or one</td><td>-</td></tr>
  <tr><td>is_anonymous</td><td>*</td><td>zero or one</td><td>-</td></tr>
  <tr><td>name</td><td>*</td><td>zero or one</td><td>names are recommended for all frames</td></tr>
  <tr><td>namespace</td><td>*</td><td>one</td><td>-</td></tr>
  <tr><td>def</td><td>*</td><td>zero or one</td><td>-</td></tr>
  <tr><td>comment</td><td>*</td><td>zero or one</td><td>-</td></tr>
  <tr><td>domain</td><td>Typedef</td><td>zero or one</td><td>use union constructs in place of multiple values</td></tr>
  <tr><td>range</td><td>Typedef</td><td>zero or one</td><td>use union constructs in place of multiple values</td></tr>
  <tr><td>is_anti_symmetric</td><td>*</td><td>zero or one</td><td>-</td></tr>
  <tr><td>is_cyclic</td><td>*</td><td>zero or one</td><td>-</td></tr>
  <tr><td>is_reflexive</td><td>*</td><td>zero or one</td><td>-</td></tr>
  <tr><td>is_symmetric</td><td>*</td><td>zero or one</td><td>-</td></tr>
  <tr><td>is_transitive</td><td>*</td><td>zero or one</td><td>-</td></tr>
  <tr><td>is_functional</td><td>*</td><td>zero or one</td><td>-</td></tr>
  <tr><td>is_inverse_functional</td><td>*</td><td>zero or one</td><td>-</td></tr>
  <tr><td>is_obsolete</td><td>*</td><td>zero or one</td><td>-</td></tr>
  <tr><td>instance_of</td><td>Instance</td><td>zero or one</td><td>use intersection constructs in place of multiple values</td></tr>
  <tr><td>intersection_of</td><td>Term or Typedef</td><td>zero or (two or more)</td><td>-</td></tr>
  <tr><td>union_of</td><td>Term or Typedef</td><td>zero or (two or more)</td><td>-</td></tr>
  <tr><td>created_by</td><td>*</td><td>zero or one</td><td>-</td></tr>
  <tr><td>creation_date</td><td>*</td><td>zero or one</td><td>-</td></tr>
  <tr><td>is_metadata_tag</td><td>Typedef</td><td>zero or one</td><td>-</td></tr>
  <tr><td>is_class_level</td><td>Typedef</td><td>zero or one</td><td>-</td></tr>

</tbody>
</table>

<ul>
  <li>
    If a frame with an intersection_of clause has exactly one intersection_of(Class-ID) clause, then the frame is said to have a <i>genus-differentia definition</i>
  </li>
</ul>

<h2><a name="5"></a>5 OBO Semantics : Mapping to OWL2-DL</h2>

<p>On completions this section will define the semantics of the
entirety of OBO via mappings to OWL2. The mappings could also be used
to specify a translation procedure and/or an interface to OWL tools
(such as OWL reasoners).</p>


<p>The translation is defined using a translation function <span
class="name">T</span> which translates (a fragment of) OBO into OWL
DL. The definition of <span class="name">T</span> is often recursive,
but it will eventually "ground out" in (a fragment of) OWL DL.</p>


<h3><a name="5.0"></a>5.0 Ontologies</h3>

<p>
  An OBO-Doc translates to an owl Ontology. The Ontology IRI is
  generated by extracting the value of the "ontology" tag in the header frame
</p>

<h3><a name="5.0.1"></a>5.0.1 Ontology Documents</h3>

<table border="1" cellspacing="0">
<caption><a id="decl" name="decl" ></a>
	Mapping of Ontology Document</caption>

<thead>
<tr><th>OBO</th>
   <th>Translation - <span class="name">T</span>(S)</th></tr>
</thead>

<tbody>

<tr>
<td valign="top" style="white-space: pre;">OBO-Doc(Header1...HeaderN Frame1..FrameM)
</td>
<td valign="top" style="white-space: pre;">Ontology(<span class="name">T<sup>OntID</sup></span>(Header[@ontology],Header[@data-version])
    <span class="name">T<sup>H</sup></span>(Header-1)...<span class="name">T<sup>H</sup></span>(Header-1)
    <span class="name">T<sup>F</sup></span>(Frame1)...<span class="name">T<sup>F</sup></span>(FrameM) )
</td>
</tr>
</tbody>
</table>


<h3><a name="5.0.2"></a>5.0.2 Ontology IRIs</h3>

<p>
An OBO ontology ID can be either an entire IRI or an abbreviated form
such as 'go'. The abbreviated form is translated by prefixing the
standard obolibrary PURL scheme and affixing with ".owl"
</p>

<p>
<span class="name">T<sup>OntID</sup></span>(IRI) = IRI
</p>
<p>
<span class="name">T<sup>OntID</sup></span>(abbreviated-ID) =
'http://purl.obolibrary.org/obo/' abbreviated-ID '.owl'
</p>

<p>
An ontology is in abbreviated form if it contains only alphanumeric
characters, plus '_', '-' and '.'
</p>

<h3><a name="5.0.3"></a>5.0.3 Ontology Headers</h3>

<p></p>


<table border="1" cellspacing="0">
<caption><a id="decl" name="decl" ></a>
  Mapping of Ontology Headers</caption>

<thead>
<tr><th>OBO</th>
   <th>Translation - <span class="name">T<sup>H</sup></span>(S)</th></tr>
</thead>

<tbody>

<tr>
<td valign="top" style="white-space: pre;">data-version(String)
</td>
<td valign="top" style="white-space: pre;">-
</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">import(IRI)
</td>
<td valign="top" style="white-space: pre;">Import(IRI)</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">remark(String)
</td>
<td valign="top" style="white-space: pre;">Annotation(<span class="name">T</span>(remark) <span class="name">T</span>(String))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">subsetdef(ID String)
</td>
<td valign="top" style="white-space: pre; color: red">AnnotationProperty(<span class="name">T</span>(ID))
SubAnnotationPropertyOf( <span class="name">T</span>(ID) <span class="name">T</span>(subsetdef) )
AnnotationAssertion(<span class="name">T</span>(name) <span class="name">T</span>(ID) ID)
AnnotationAssertion(<span class="name">T</span>(comment) <span class="name">T</span>(ID) String)
</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">synonymtypedef(ID String [Scope])
</td>
<td valign="top" style="white-space: pre; color:red">AnnotationProperty(<span class="name">T</span>(ID))
SubAnnotationPropertyOf( <span class="name">T</span>(ID) <span class="name">T</span>(subsetdef) )
AnnotationAssertion(<span class="name">T</span>(name) <span class="name">T</span>(ID) <span class="name">T</span>(String))
[AnnotationAssertion(<span class="name">T</span>(hasScope) <span class="name">T</span>(ID) <span class="name">T</span>(Scope))]
</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">&lt;Tag&gt;(String Qualifiers)</td>
<td valign="top" style="white-space: pre;">AnnotationAssertion(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(&lt;Tag&gt;) <span class="name">T</span>(ID) <span class="name">T</span>(String))</td>
</tr>

</tbody>
</table>

<h3><a name="5.0.4"></a>5.0.4 Untranslatable OWL axioms</h3>

<p style="color: red;">
An ontology can contain any number of arbitrary OWL axioms encoded in
OWL functional syntax in the ontology header using the owl-axioms
tag. The values of all owl-axioms clauses are concatenated together
(with newline characters appended) in
order of appearance in the document and interpreted as if they were
enclosed within an "Ontology( ... )" production in functional
syntax. This allows any OWL ontology to be roundtripped through
OBO-format, albeit in such a way that expressive OWL axioms are opaque
to many OBO-format applications. 
</p>



<h4 style="color: red;">Prefixes</h4>

<p style="color: red;">
The following prefixes are assumed to be pre-declared - other prefixes
cannot be declared and thus full IRIs must be used

<span style="white-space: pre; color: red;">
Prefix: xsd: &lt;http://www.w3.org/2001/XMLSchema#&gt;
Prefix: owl: &lt;http://www.w3.org/2002/07/owl#&gt;
Prefix: : &lt;http://purl.obolibrary.org/obo/&gt;
Prefix: oboInOwl: &lt;http://www.geneontology.org/formats/oboInOwl#&gt;
Prefix: xml: &lt;http://www.w3.org/XML/1998/namespace&gt;
Prefix: rdf: &lt;http://www.w3.org/1999/02/22-rdf-syntax-ns#&gt;
Prefix: dc: &lt;http://purl.org/dc/elements/1.1/&gt;
Prefix: rdfs: &lt;http://www.w3.org/2000/01/rdf-schema#&gt;
</span>

It is recommended that writers use this set of prefixes to write
CURIES (shortened URIs), but this is not required.

</p>

<h4 style="color: red;">Example</h4>

<p style="white-space: pre; color: red;">
format-version: 1.2
ontology: go
date: 17:04:2012 15:38
...
remark: this is an example of what a header in the GO might look like
remark: We have two axioms, the first declares that nothing is part of
remark: both a nucleus and a cytoplasm, the second declares that
remark: nothing is part of both a cytoplasm and a plasma membrane.
remark: (Note we could have chosen to use a single axiom with 3
remark: arguments, which is better as it also gives us the spatial
remark: disjointness between nucleus and PM)
owl-axioms: DisjointClasses(ObjectSomeValuesFrom(:BFO_0000050 :GO_0005737) ObjectSomeValuesFrom(:BFO_0000050 :GO_0005634))
owl-axioms: DisjointClasses(ObjectSomeValuesFrom(:BFO_0000050 :GO_0005737) ObjectSomeValuesFrom(:BFO_0000050 :GO_0005886))

[Term]
...
</p>

<h4 style="color: red;">Behavior for OBO-writers (normative)</h4>

<p style="color: red;">
When translating an OWL ontology to an OBO-document, any axiom that is
not translated according to the standard obo2owl rules must be added
as an owl-axioms clause
</p>

<h4 style="color: red;">Recommendations for OBO-writers (informative)</h4>

<div style="color: red;">
Note that individual axioms may be broken over multiple lines;
alternatively, multiple axioms may be concatenated as one line. It is
recommended that writers write each axiom on its own line/clause.

<h5 style="color: red;">Recommended</h5>
<span style="white-space: pre; color: red;">
owl-axioms: DisjointClasses(ObjectSomeValuesFrom(:BFO_0000050 :GO_0005737) ObjectSomeValuesFrom(:BFO_0000050 :GO_0005634))
owl-axioms: DisjointClasses(ObjectSomeValuesFrom(:BFO_0000050 :GO_0005737) ObjectSomeValuesFrom(:BFO_0000050 :GO_0005886))
</span>

<h5 style="color: red;">Allowed</h5>
<span style="white-space: pre; color: red;">
owl-axioms: DisjointClasses(ObjectSomeValuesFrom(:BFO_0000050 :GO_0005737) 
owl-axioms:                 ObjectSomeValuesFrom(:BFO_0000050 :GO_0005634))
owl-axioms: DisjointClasses(ObjectSomeValuesFrom(:BFO_0000050 :GO_0005737) 
owl-axioms:                 ObjectSomeValuesFrom(:BFO_0000050 :GO_0005886))
</span>

<h5 style="color: red;">Allowed</h5>
<span style="white-space: pre; color: red;">
owl-axioms: DisjointClasses(ObjectSomeValuesFrom(:BFO_0000050 :GO_0005737) ObjectSomeValuesFrom(:BFO_0000050 :GO_0005634)) DisjointClasses(ObjectSomeValuesFrom(:BFO_0000050 :GO_0005737) ObjectSomeValuesFrom(:BFO_0000050 :GO_0005886))
</span>

</div>



<p style="color: red;">
It is strongly recommended that writers only generate owl-axioms
clauses for axioms that fall outside of the OBO-subset of
OWL2-DL. Remember that most applications which consume obo-format will
ignore the owl-axioms headers.
</p>



<h3><a name="5.1"></a>5.1 Declarations</h3>

<table border="1" cellspacing="0">
<caption><a id="decl" name="decl" ></a>
	Mapping of frames/stanzas to OWL</caption>

<thead>
<tr><th>OBO</th>
  <th>Conditions</th>
  <th>Translation - <span class="name">T<sup>F</sup></span>(S)</th></tr>
</thead>

<tbody>

<tr>
<td valign="top" style="white-space: pre;">Term(Class-id Clause-1..Clause-n)
</td>
<td>
</td>
<td valign="top" style="white-space: pre;">Class(<span class="name">T</span>(Class-id)) <span class="name">T<sup>C</sup></span>(Clause-1)..<span class="name">T<sup>C</sup></span>(Clause-n)
</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">Typedef(Rel-id)
</td>
<td>is_metadata_tag(true)
</td>
<td valign="top" style="white-space: pre;">AnnotationProperty(<span class="name">T</span>(Rel-id)) <span class="name">T<sup>P</sup></span>(Clause-1)..<span class="name">P<sup>C</sup></span>(Clause-n)
</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">Typedef(Rel-id)
</td>
<td>NOT (is_metadata_tag(true))
</td>

<td valign="top" style="white-space: pre;">ObjectProperty(<span class="name">T</span>(Rel-id)) <span class="name">T<sup>P</sup></span>(Clause-1)..<span class="name">P<sup>C</sup></span>(Clause-n)
</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">Instance(Instance-id)
</td>
<td>
</td>
<td valign="top" style="white-space: pre;">NamedIndividual(<span class="name">T</span>(Instance-id)) <span class="name">T<sup>I</sup></span>(Clause-1)..<span class="name">I<sup>C</sup></span>(Clause-n)
</td>
</tr>
</tbody>

</table>

<h3><a name="5.2"></a>5.2 Mapping Term Frames - Class Axioms</h3>
<h4><a name="5.2.1"></a>5.2.1 Class Axioms, Basic Table</h4>

<p>
  Term frames are mapped to Class declarations and Class-level
  axioms. We list the translations that are specific to Term frames
  here - generic translations making annotation assertions are applied
  later on.
</p>

<table border="1" cellspacing="0">
<caption><a id="translation" name="translationDL" />
	Translation of Term frames to OWL</caption>

<thead>
<tr><th>Term frame with id:Class-ID</th>
  <th>Translation - <span class="name">T<sup>C</sup></span>(S)</th></tr>
</thead>

<tbody>

<tr>
<td valign="top" style="white-space: pre;">is_a(SubClass-ID Qualifiers)
</td>
<td valign="top" style="white-space: pre;">SubClassOf(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Class-ID) <span class="name">T</span>(SubClass-ID))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">relationship(Rel-ID TargetClass-ID Qualifiers) <br/><i>on condition:</i> is_class_level(true) &isin; TypedefFrames[Rel-ID]
</td>
<td valign="top" style="white-space: pre;">AnnotationAssertion(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Class-ID) <span class="name">T</span>(Rel-ID) <span class="name">T</span>(Target-Class-ID) )</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">relationship(Rel-ID TargetClass-ID Qualifiers) <br/><i>on condition:</i> is_class_level(true) &notin; TypedefFrames[Rel-ID]
</td>
<td valign="top" style="white-space: pre;">SubClassOf(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Class-ID) <span class="name">T</span>(rel(Rel-ID TargetClass-ID Qualifiers)) )</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">intersection_of(GenusClass1 Qualifiers1) 
... 
intersection_of(GenusClassN QualifiersN) 
...
intersection_of(Rel-ID1 TargetClass-ID1 QualifiersN+1)
...
intersection_of(Rel-IDM TargetClass-IDM QualifiersN+M)
</td>
<td valign="top" style="white-space: pre;">EquivalentClasses(<span class="name">T</span>(Qualifiers1 .. QualifiersN+M)
    <span class="name">T</span>(Class-ID) 
    ObjectIntersectionOf(
       <span class="name">T</span>(GenusClass1) 
       ... 
       <span class="name">T</span>(GenusClassN)
       <span class="name">T</span>(rel(Rel-ID1 TargetClass-ID1 Qualifiers))
       ... 
       <span class="name">T</span>(rel(Rel-IDM TargetClass-IDM Qualifiers)))</td>
</tr>
<tr>
<td valign="top" style="white-space: pre;">union_of(UnionClass1 Qualifiers1)
... 
union_of(UnionClassN QualifiersN) 
</td>
<td valign="top" style="white-space: pre;">EquivalentClasses(<span class="name">T</span>(Qualifiers1 .. QualifiersN)
    <span class="name">T</span>(Class-ID) 
    ObjectUnionOf(
       <span class="name">T</span>(UnionClass1) ... <span class="name">T</span>(UnionClassN) )
</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">disjoint_from(TargetClass-ID Qualifiers))
</td>
<td valign="top" style="white-space: pre;">DisjointClasses(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Class-ID) <span class="name">T</span>(TargetClass-ID))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">equivalent_to(TargetClass-ID Qualifiers))
</td>
<td valign="top" style="white-space: pre;">EquivalentClasses(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Class-ID) <span class="name">T</span>(TargetClass-ID))</td>
</tr>

</tbody>
</table>

<h4><a name="5.2.2"></a>5.2.2 Treatment of gci_relation qualifier</h4>

              <span style="color: red;">
                <p>
                  If <span style="white-space: pre;">Qualifiers</span> contains the tags <span style="white-space: pre;">gci_relation</span> and  <span style="white-space: pre;">gci_filler</span> then
                  <span style="white-space:pre;"><span class="name">T</span>(Class-ID)</span> in 5.2.1 is translated as a class expression:
                </p>
                <span style="white-space: pre;">ObjectIntersectionOf(<span class="name">T</span>(Class-ID) ObjectSomeValuesFrom(<span class="name">T</span>(GCI-Relation) <span class="name">T</span>(GCI-Filler)))</span>
                <p>
                </p>

                <p>
                  This effectively allows a limited form of General Class Inclusion (GCI) axioms in OBO Documents. For example, in:
                  <span style="white-space: pre;"> 

id: CL:0000232
name: erythrocyte
relationship: has_part GO:0005634 {gci_relation="part_of", gci_filler="NCBITaxon:7955"} ! nucleus

                  </span>

                  the relationship is translated as:

                  <span style="white-space: pre;"> 

SubClassOf(
   ObjectIntersectionOf( ObjectSomeValuesFrom(T('part_of') T('NCBITaxon:7955')) T('CL:0000232'))
   ObjectSomeValuesFrom(T('has_part') T('GO:0005634'))
)

                  </span>

The resulting axiom would be (in Manchester-like syntax, informative):

                  <span style="white-space: pre;"> 

(erythrocyte and part_of some 'Danio rerio') SubClassOf has_part some nucleus

                  </span>


                </p>
              </span>

<h3><a name="5.3"></a>5.3 Class Expressions</h3>

<p>
  The relationship and intersection_of tags can take pairs of values;
  these pairs map to OWL Class Expressions according to the following
  table.
</p>


<p>
  Mappings are listed in order of precedence - an expression cannot
  match more than one production rule, only the first match counts.
</p>

<table border="1" cellspacing="0">
<caption><a id="translation" name="translationDL" ></a>
	Translation of class expressions to OWL</caption>

<thead>
<tr><th>Expression</th>
    <th>Condition</th>
    <th>Translation - <span class="name">T</span>(S)</th></tr>
</thead>

<tbody>

<tr>
<td valign="top" style="white-space: pre;">rel(Rel-ID Class-ID Qualifiers)
</td>
<td valign="top" style="white-space: pre;">cardinality(X) &isin; Qualifiers
</td>
<td valign="top" style="white-space: pre;">ObjectExactCardinality(X <span class="name">T</span>(Rel-ID) <span class="name">T</span>(Class-ID))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">rel(Rel-ID Class-ID Qualifiers)
</td>
<td valign="top" style="white-space: pre;">cardinality(0) &isin; Qualifiers or
maxCardinality(0) &isin; Qualifiers
</td>
<td valign="top" style="white-space: pre;">ObjectAllValuesFrom(<span class="name">T</span>(Rel-ID) ObjectComplementOf(<span class="name">T</span>(Class-ID)))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">rel(Rel-ID Class-ID Qualifiers)
</td>
<td valign="top" style="white-space: pre;">minCardinality(A) &isin; Qualifiers and
maxCardinality(B) &isin; Qualifiers
</td>
<td valign="top" style="white-space: pre;">ObjectIntersectionOf(
    ObjectMinCardinality(A <span class="name">T</span>(Rel-ID) <span class="name">T</span>(Class-ID))
    ObjectMaxCardinality(B <span class="name">T</span>(Rel-ID) <span class="name">T</span>(Class-ID)))
</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">rel(Rel-ID Class-ID Qualifiers)</td>
<td valign="top" style="white-space: pre;">minCardinality(A) &isin; Qualifiers</td>
<td valign="top" style="white-space: pre;">ObjectMinCardinality(A <span class="name">T</span>(Rel-ID) <span class="name">T</span>(Class-ID))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">rel(Rel-ID Class-ID Qualifiers)</td>
<td valign="top" style="white-space: pre;">maxCardinality(A) &isin; Qualifiers</td>
<td valign="top" style="white-space: pre;">ObjectMaxCardinality(A <span class="name">T</span>(Rel-ID) <span class="name">T</span>(Class-ID))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">rel(Rel-ID Class-ID Qualifiers)
</td>
<td valign="top" style="white-space: pre;">all_only(true) &isin; Qualifiers and
all_some(true) &isin; Qualifiers
</td>
<td valign="top" style="white-space: pre;">ObjectIntersectionOf(
  ObjectSomeValuesFrom(X <span class="name">T</span>(Rel-ID) <span class="name">T</span>(Class-ID)
  ObjectAllValuesFrom(X <span class="name">T</span>(Rel-ID) <span class="name">T</span>(Class-ID)
)</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">rel(Rel-ID Class-ID Qualifiers)
</td>
<td valign="top" style="white-space: pre;">all_only(true) &isin; Qualifiers
</td>
<td valign="top" style="white-space: pre;">ObjectAllValuesFrom(<span class="name">T</span>(Rel-ID) <span class="name">T</span>(Class-ID))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">rel(Rel-ID Class-ID Qualifiers)
</td>
<td valign="top" style="white-space: pre;">is_class_level(true) &isin; TypedefFrames[Rel-ID]
</td>
<td valign="top" style="white-space: pre;">ObjectHasValue(<span class="name">T</span>(Rel-ID) <span class="name">T</span>(Class-ID))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">rel(Rel-ID Class-ID Qualifiers)
</td>
<td></td>
<td valign="top" style="white-space: pre;">ObjectSomeValuesFrom(<span class="name">T</span>(Rel-ID) <span class="name">T</span>(Class-ID)</td>
</tr>


</tbody>
</table>

<h3><a name="5.4"></a>5.4 Property Axioms</h3>

<p>
  Note that OWL2 does not have the ability to declare
  EquivalentProperties between a named object property and either an
  intersection or a union. We translate these OBO constructs to a
  weaker axiom and add an AnnotationAssertion - we will later provide
  an appendix for handling these in FOL. This is also true for equivalent_to_chain
</p>

<table border="1" cellspacing="0">
<caption><a id="translation" name="translationDL" ></a>
	Translation of Typedef frames to OWL</caption>

<thead>
<tr><th>Typedef frame with id:Rel-ID</th>
    <th>Translation - <span class="name">T<sup>P</sup></span>(S)</th></tr>
</thead>

<tbody>

<tr>
<td valign="top" style="white-space: pre;">is_a(SubRel-ID Qualifiers))
</td>
<td valign="top" style="white-space: pre;">SubObjectPropertyOf(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Rel-ID) <span class="name">T</span>(SubRel-ID))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">domain(TargetClass-ID Qualifiers))
</td>
<td valign="top" style="white-space: pre;">ObjectPropertyDomain(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Rel-ID) <span class="name">T</span>(TargetClass-ID))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">range(TargetClass-ID Qualifiers))
</td>
<td valign="top" style="white-space: pre;">ObjectPropertyRange(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Rel-ID) <span class="name">T</span>(TargetClass-ID))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">disjoint_from(TargetRel-ID Qualifiers))
</td>
<td valign="top" style="white-space: pre;">DisjointObjectProperties(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Rel-ID) <span class="name">T</span>(TargetRel-ID))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">inverse_of(TargetRel-ID Qualifiers))
</td>
<td valign="top" style="white-space: pre;">InverseProperties(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Rel-ID) <span class="name">T</span>(TargetRel-ID))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">equivalent_to(TargetRel-ID Qualifiers))
</td>
<td valign="top" style="white-space: pre;">EquivalentObjectProperties(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Rel-ID) <span class="name">T</span>(TargetRel-ID))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">relationship(MetaRel-ID TargetRel-ID Qualifiers)
</td>
<td valign="top" style="white-space: pre;">AnnotationAssertion(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(MetaRel-ID) <span class="name">T</span>(Rel-ID) <span class="name">T</span>(TargetRel-ID) )</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">intersection_of(TargetRel-ID Qualifiers)</td>
<td valign="top" style="white-space: pre;">SubObjectPropertyOf(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Rel-ID) (TargetRel-ID))
AnnotationAssertion(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(relation_intersection_of) <span class="name">T</span>(Rel-ID) <span class="name">T</span>(TargetRel-ID))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">union_of(TargetRel-ID Qualifiers)</td>
<td valign="top" style="white-space: pre;">SubObjectPropertyOf(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(TargetRel-ID) <span class="name">T</span>(Rel-ID))
AnnotationAssertion(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(union_intersection_of) <span class="name">T</span>(Rel-ID) <span class="name">T</span>(TargetRel-ID))</td>
</tr>
<tr>

<tr>
<td valign="top" style="white-space: pre;">transitive_over(Rel2-ID Qualifiers)</td>
<td valign="top" style="white-space: pre;">SubObjectPropertyOf(<span class="name">T</span>(Qualifiers) ObjectPropertyChain( <span class="name">T</span>(Rel-ID) <span class="name">T</span>(Rel2-ID) ) <span class="name">T</span>(Rel-ID) )</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">holds_over_chain(Rel1-ID Rel2-ID Qualifiers)</td>
<td valign="top" style="white-space: pre;">SubObjectPropertyOf(<span class="name">T</span>(Qualifiers) ObjectPropertyChain( <span class="name">T</span>(Rel1-ID) <span class="name">T</span>(Rel2-ID) ) <span class="name">T</span>(Rel-ID) )</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">equivalent_to_chain(Rel1-ID Rel2-ID Qualifiers)</td>
<td valign="top" style="white-space: pre;">SubObjectPropertyOf(<span class="name">T</span>(Qualifiers) ObjectPropertyChain( <span class="name">T</span>(Rel1-ID) <span class="name">T</span>(Rel2-ID) ) <span class="name">T</span>(Rel-ID) )
 AnnotationAssertion( <span style="color: red">TODO</span> )</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">disjoint_over(OverRel-ID Qualifiers)</td>
<td valign="top" style="white-space: pre;">AnnotationAssertion(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(disjoint_over) <span class="name">T</span>(Rel-ID) <span class="name">T</span>(OverRel-ID))</td>
</tr>
<tr>

<tr>
<td valign="top" style="white-space: pre;">is_cyclic(true Qualifiers)</td>
<td valign="top" style="white-space: pre;">AnnotationAssertion(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(is_cyclic) <span class="name">T</span>(Rel-ID) <span class="name">T</span>(true))</td>
</tr>
<tr>


<tr>
<td valign="top" style="white-space: pre;">is_transitive(true Qualifiers)</td>
<td valign="top" style="white-space: pre;">TransitiveObjectProperty(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Rel-ID))</td>
</tr>
<tr>

<tr>
<td valign="top" style="white-space: pre;">is_anti_symmetric(true Qualifiers)</td>
<td valign="top" style="white-space: pre;">AnnotationAssertion(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(is_anti_symmetric) <span class="name">T</span>(Rel-ID) <span class="name">T</span>(true))</td>
</tr>
<tr>

<tr>
<td valign="top" style="white-space: pre;">is_reflexive(true Qualifiers)</td>
<td valign="top" style="white-space: pre;">ReflexiveObjectProperty(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Rel-ID))</td>
</tr>
<tr>

<tr>
<td valign="top" style="white-space: pre;">is_symmetric(true Qualifiers)</td>
<td valign="top" style="white-space: pre;">SymmetricObjectProperty(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Rel-ID))</td>
</tr>
<tr>

<tr>
<td valign="top" style="white-space: pre;">is_asymmetric(true Qualifiers)</td>
<td valign="top" style="white-space: pre;">AsymmetricObjectProperty(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Rel-ID))</td>
</tr>
<tr>

<tr>
<td valign="top" style="white-space: pre;">is_functional(true Qualifiers)</td>
<td valign="top" style="white-space: pre;">FunctionalObjectProperty(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Rel-ID))</td>
</tr>
<tr>

<tr>
<td valign="top" style="white-space: pre;">is_inverse_functional(true Qualifiers)</td>
<td valign="top" style="white-space: pre;">InverseFunctionalObjectProperty(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Rel-ID))</td>
</tr>
<tr>

<tr>
<td valign="top" style="white-space: pre;">expand_expression_to(String Xrefs Qualifiers)</td>
<td valign="top" style="white-space: pre;">AnnotationAssertion(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(expand_expression_to) <span class="name">T</span>(Rel-ID) <span class="name">T</span>(String))</td>
</tr>
<tr>
<tr>
<td valign="top" style="white-space: pre;">expand_assertion_to(String Xrefs Qualifiers)</td>
<td valign="top" style="white-space: pre;">AnnotationAssertion(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(expand_assertion_to) <span class="name">T</span>(Rel-ID) <span class="name">T</span>(String))</td>
</tr>

</tbody>
</table>

<h4>Inferred Axioms</h4>

              <p>
                The OBO Format equivalent_to_chain clause can only be
                partially expressed in OWL2. We can do some inference
                as part of translation -
                see <a href="https://github.com/cmungall/obo2owl/blob/master/hoc.p9">this
                prover9 file</a> for a proof.
              </p>
              
<table border="1" cellspacing="0">
<caption><a id="translation-i" name="translationI" ></a>
	Additional axioms generated by equivalent_to_chain clauses</caption>

<thead>
<tr><th>Typedef frame with id:Rel-ID</th>
    <th>Translation - <span class="name">T<sup>P</sup></span>(S)</th></tr>
</thead>
<tr>
<td valign="top" style="white-space: pre;">equivalent_to_chain(Rel1-ID Rel2-ID Qualifiers)
is_transitive(Rel2-ID)</td>
<td valign="top" style="white-space: pre;">SubObjectPropertyOf(ObjectPropertyChain( <span class="name">T</span>(Rel-ID) <span class="name">T</span>(Rel2-ID) ) <span class="name">T</span>(Rel-ID) )</td>
</tr>
<tr>
<td valign="top" style="white-space: pre;">equivalent_to_chain(Rel1-ID Rel2-ID Qualifiers)
is_transitive(Rel1-ID)</td>
<td valign="top" style="white-space: pre;">SubObjectPropertyOf(ObjectPropertyChain( <span class="name">T</span>(Rel-ID) <span class="name">T</span>(Rel1-ID) ) <span class="name">T</span>(Rel-ID) )</td>
</tr>
</table>

<h3><a name="5.5"></a>5.5 Individual Axioms</h3>

<table border="1" cellspacing="0">
<caption><a id="translation-i" name="translationI" ></a>
	Translation of Instance frames to OWL</caption>

<thead>
<tr><th>Instance frame with id:Instance-ID</th>
    <th>Translation - <span class="name">T</span>(S)</th></tr>
</thead>

<tbody>

<tr>
<td valign="top" style="white-space: pre;">instance_of(Class-ID Qualifiers)</td>
<td valign="top" style="white-space: pre;">ClassAssertion(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Class-ID) <span class="name">T</span>(Instance-ID))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">relationship(Rel-ID ID)</td>
<td valign="top" style="white-space: pre;">PropertyAssertion(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Rel-ID) <span class="name">T</span>(Instance-ID) <span class="name">T</span>(ID))</td>
</tr>

</tbody>
</table>


<h3><a name="5.6"></a>5.6 Common Elements to Annotation Properties</h3>

<table border="1" cellspacing="0">
<caption><a id="translation-ap" name="translationAP" ></a>
	Translation of common frame elements</caption>

<thead>
<tr><th>Frame with id:ID</th>
    <th>Translation - <span class="name">T</span>(S)</th></tr>
</thead>

<tbody>

<tr>
<td valign="top" style="white-space: pre;">name(String Qualifiers)</td>
<td valign="top" style="white-space: pre;">AnnotationAssertion(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(name) <span class="name">T</span>(ID) <span class="name">T</span>(String))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">def(String Xrefs Qualifiers)</td>
<td valign="top" style="white-space: pre;">AnnotationAssertion(<span class="name">T</span>(ann(Xrefs Qualifiers)) <span class="name">T</span>(def) <span class="name">T</span>(ID) <span class="name">T</span>(String))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">synonym(String 'EXACT' [Type] Xrefs Qualifiers)</td>
<td valign="top" style="white-space: pre; color:red">AnnotationAssertion(<span class="name">T</span>(ann( [Type] Xrefs Qualifiers)) <span class="name">T</span>(exactSynonyn) <span class="name">T</span>(ID) <span class="name">T</span>(String))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">synonym(String 'NARROW' [Type] Xrefs Qualifiers)</td>
<td valign="top" style="white-space: pre; color:red">AnnotationAssertion(<span class="name">T</span>(ann( [Type] Xrefs Qualifiers)) <span class="name">T</span>(narrowSynonym) <span class="name">T</span>(ID) <span class="name">T</span>(String))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">synonym(String 'RELATED' [Type] Xrefs Qualifiers)</td>
<td valign="top" style="white-space: pre; color:red">AnnotationAssertion(<span class="name">T</span>(ann( [Type] Xrefs Qualifiers)) <span class="name">T</span>(relatedSynonym) <span class="name">T</span>(ID) <span class="name">T</span>(String))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">synonym(String 'BROAD' [Type] Xrefs Qualifiers)</td>
<td valign="top" style="white-space: pre; color:red">AnnotationAssertion(<span class="name">T</span>(ann( [Type] Xrefs Qualifiers)) <span class="name">T</span>(broadSynonym) <span class="name">T</span>(ID) <span class="name">T</span>(String))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">creation_date(ISO-Date Qualifiers)</td>
<td valign="top" style="white-space: pre;">AnnotationAssertion(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(creation_date) <span class="name">T</span>(ID) <span class="name">T</span>(ISO-Date))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">xref(X-ID Qualifiers)</td>
<td valign="top" style="white-space: pre;">AnnotationAssertion(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(xref) <span class="name">T</span>(ID) <span class="name">T</span>(X-ID))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">property_value(Rel-ID Entity-ID Qualifiers)</td>
<td valign="top" style="white-space: pre;">AnnotationAssertion(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Rel-ID) <span class="name">T</span>(ID) <span class="name">T</span>(Entity-ID))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">property_value(Rel-ID Value XSD-Type Qualifiers)</td>
<td valign="top" style="white-space: pre;">AnnotationAssertion(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(Rel-ID) <span class="name">T</span>(ID) <span class="name">T<sup>literal</sup></span>(Value XSD-Type))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">subset(Subsetdef-ID)</td>
<td valign="top" style="white-space: pre; color:red">AnnotationAssertion(<span class="name">T</span>(subset) <span class="name">T</span>(ID) <span class="name">T</span>(Subsetdef-ID))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">&lt;Tag&gt;(String Qualifiers)</td>
<td valign="top" style="white-space: pre;">AnnotationAssertion(<span class="name">T</span>(Qualifiers) <span class="name">T</span>(&lt;Tag&gt;) <span class="name">T</span>(ID) <span class="name">T</span>(String))</td>
</tr>

</tbody>
</table>

<h3><a name="5.7"></a>5.7 Translation of OBO Qualifiers to Annotations</h3>

<table border="1" cellspacing="0">
<caption><a id="translation-q" name="translationQ" ></a>
	Translation of qualifier expressions</caption>

<thead>
<tr><th>Expression</th>
    <th>Translation - <span class="name">T</span>(S)</th></tr>
</thead>

<tbody>

<tr>
<td valign="top" style="white-space: pre;">Qualifier1 ... QualifierN</td>
<td valign="top" style="white-space: pre;"><span class="name">T</span>(Qualifier1) ... <span class="name">T</span>(QualifierN)</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">Qualifier(Name,Value)</td>
<td valign="top" style="white-space: pre;">Annotation( <span class="name">T</span>(Name) <span class="name">T</span>(Value))</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">ann(Xref1...XrefN Qualifiers)</td>
<td valign="top" style="white-space: pre;">Annotation( <span class="name">T</span>(xref) <span class="name">T</span>(Xref1)) .. Annotation( <span class="name">T</span>(xref) <span class="name">T</span>(Xref1)) <span class="name">T</span>(Qualifiers)</td>
</tr>

<!--
<tr>
<td valign="top" style="white-space: pre;">ann( Xrefs Qualifiers)</td>
<td valign="top" style="white-space: pre;">Annotation( <span class="name">T</span>(synonym-scope) <span class="name">T</span>(Scope)) <span class="name">T</span>(Xrefs Qualifiers)</td>
</tr>
-->

<tr>
<td valign="top" style="white-space: pre;">ann( Type Xrefs Qualifiers)</td>
<td valign="top" style="white-space: pre; color:red">Annotation( <span class="name">T</span>(synonym-type) <span class="name">T</span>(Type)) <span class="name">T</span>(Scope Xrefs Qualifiers)</td>
</tr>

</tbody>

</table>

<div>
The following qualifiers are not translated to annotations:
<ul>
 <li>cardinality</li>
 <li>maxCardinality</li>
 <li>maxCardinality</li>
 <li>gci_relation</li>
 <li>gci_filler</li>
</ul>
</div>

<h3><a name="5.8"></a>5.8 Translation of Annotation Vocabulary</h3>

<div>
 Annotation properties are drawn from the followling vocabularies:
                <ul>
                  <li>
                    IAO  - http://purl.obolibrary.org/obo/IAO_
                  </li>
                  <li>
                    oboInOwl  - http://www.geneontology.org/formats/oboInOwl#
                  </li>
                  <li>
                    rdfs  - http://www.w3.org/2000/01/rdf-schema#
                  </li>
                  <li>
                    dc  - http://purl.org/dc/elements/1.1/
                  </li>

                </ul>
</div>


<table border="1" cellspacing="0">
<caption><a id="translation-q" name="translationQ" ></a>
	Translation of identifiers - hardcoded values</caption>

<thead>
<tr><th>ID</th>
    <th>Translation - <span class="name">T</span>(S)</th></tr>
</thead>

<tbody>

<tr>
<td valign="top" style="white-space: pre;">is_obsolete</td>
<td valign="top" style="white-space: pre;">owl:deprecated</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">name</td>
<td valign="top" style="white-space: pre;">rdfs:label</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">comment</td>
<td valign="top" style="white-space: pre;">rdfs:comment</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">date</td>
<td valign="top" style="white-space: pre;">dc:date</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">expand_expression_to</td>
<td valign="top" style="white-space: pre;">obo:IAO_0000424</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">expand_assertion_to</td>
<td valign="top" style="white-space: pre;">obo:IAO_0000425</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">definition</td>
<td valign="top" style="white-space: pre;">obo:IAO_0000115</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">is_anti_symmetric</td>
<td valign="top" style="white-space: pre;">obo:IAO_0000427</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">replaced_by</td>
<td valign="top" style="white-space: pre;">obo:IAO_0100001</td>
</tr>


<tr>
<td valign="top" style="white-space: pre;">consider</td>
<td valign="top" style="white-space: pre; color:red">oboInOwl:consider</td>
</tr>


<tr>
<td valign="top" style="white-space: pre;">hasScope</td>
<td valign="top" style="white-space: pre; color:red">oboInOwl:hasScope</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">synonym-type</td>
<td valign="top" style="white-space: pre; color:red">oboInOwl:hasSynonymType</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">exactSynonym</td>
<td valign="top" style="white-space: pre; color:red">oboInOwl:hasExactSynonym</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">narrowSynonym</td>
<td valign="top" style="white-space: pre; color:red">oboInOwl:hasNarrowSynonym</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">broadSynonym</td>
<td valign="top" style="white-space: pre; color:red">oboInOwl:hasBroadSynonym</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">relatedSynonym</td>
<td valign="top" style="white-space: pre; color:red">oboInOwl:hasRelatedSynonym</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">synonymtypedef</td>
<td valign="top" style="white-space: pre; color:red">oboInOwl:SynonymTypeProperty</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">xref</td>
<td valign="top" style="white-space: pre; color:red">oboInOwl:xref</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">subsetdef</td>
<td valign="top" style="white-space: pre; color:red">oboInOwl:SubsetProperty</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">subset</td>
<td valign="top" style="white-space: pre; color:red">oboInOwl:inSubset</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">alt_id</td>
<td valign="top" style="white-space: pre; color:red">oboInOwl:hasAlternativeId</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">shorthand</td>
<td valign="top" style="white-space: pre; color:red">oboInOwl:shorthand</td>
</tr>

</tbody>
</table>

<p>
Note we have no cognate of IAO:definition_source - we would have a generic xref annotation on the definition annotation assertion
</p>

<h4><a name="5.8.1"></a>5.8.1 Semantics of synonyms</h4>

<p>
  Mapping of OBO Format synonym clauses is described
  in <a href="#5.6">5.6</a>. These map to annotation properties
  without any semantics in OWL. This section provides semantics
  external to OWL.
</p>

<h5>Informal Description (Informative)</h5>

<p>
In OBO Format, the term "synonym" is used loosely for any kind of
alternative label for a class (the "name" tag is used for the
community preferred label). A label is a synonym for a class if there
exists some user or user community (existing or historic) for which
this label unambiguously denotes the class.
</p>

<p>
Synonyms are always scoped into one of four disjoint categories:
EXACT, BROAD, NARROW, RELATED. A synonym is EXACT if it is a "true"
synonym - most members of the community served by the ontology regard
the exact synonym as substitutable for the primary label. If an
ontology contains two classes which share either primary label (name)
or exact synonym, then these classes are equivalent.  A synonym for a
class C is BROAD if it denotes a broader class than C. Here, broader
than is an informal notion that encompasses both subsumption and
possibly mereological and temporal containment. For example, "skull"
could conceivably be a BROAD synonym for the
class <i>cranium</i>. Conversely a synonym for a class C is NARROW if
C denotes a broader class than the synonym. If a synonym is neither
EXACT, NARROW or BROAD, then it is RELATED.
</p>

<p>
Scoping of synonyms has proven extremely useful for OBO format
ontologies - most ontologies use this feature. The ability to
designate a synonym as EXACT, in particular, is very useful for
introducing additional terminological precision and reducing ambiguity
in ontologies. These have also proven useful for NLP purposes.
</p>



<h3><a name="5.9"></a>5.9 Translation of Identifiers</h3>

<h4><a name="5.9.1"></a>5.9.1. Pre-processing</h4>
<div>Every OBO-Document is automatically populated with two clauses:
 <ul>
   <li>idspace(RO "http://purl.obolibrary.org/obo/RO_")</li>
   <li>idspace(BFO "http://purl.obolibrary.org/obo/BFO_")</li>
 </ul>
(see <a href="#4.4.1">4.4.1</a>)
</div>

<h4><a name="5.9.2"></a>5.9.2. Translation of identifiers</h4>

<p>The following table is used when translating an OBO identifier to an IRI</p>

<table border="1" cellspacing="0">
<caption><a id="translation-id" name="translationID" ></a>
	Translation of identifiers</caption>

<thead>
<tr><th>ID</th>
    <th>Translation - <span class="name">T</span>(S)</th></tr>
</thead>

<tbody>

<tr>
<td valign="top" style="white-space: pre;">Canonical-Prefixed-ID</td>
<td valign="top" style="white-space: pre;"><span class="name">T<sup>prefix</sup></span>(Canonical-IDPrefix) <b>'_' Canonical-LocalID</b></td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">NonCanonical-Prefixed-ID</td>
<td valign="top" style="white-space: pre;"><span class="name">T<sup>prefix</sup></span>(Any-IDPrefix) <b>'#_' Any-LocalID</b></td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">Unprefixed-ID</td>
<td valign="top" style="white-space: pre;">OntologyIRI <b>'#' Unprefixed-ID</b></td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">URL-as-ID</td>
<td valign="top" style="white-space: pre;"> <b>URL-as-ID</b></td>
</tr>

</tbody>
</table>

<h4><a name="5.9.3"></a>5.9.3. Special Rules for Relations</h4>

                <div>
                  An exception is made to 5.9.2 for translating relation IDs:
                  <ul>
                    <li>If an Unprefixed-ID is declared in a Typedef
                    frame, and that Typedef frame has an xref clause,
                    then the value of that xref clause is used in the
                    translation.
                    </li>
                    <li>
                      If there are multiple xref clauses in the
                      Typedef frame, then precedence is as follows:
                      <ul>
                        <li>
                          xrefs in the 'BFO' and 'RO' ID-space take
                          priority over all others
                        </li>
                        <li>
                          xrefs for which there is a corresponding
                          idspace declaration in the header of the
                          same document take precedence over xrefs
                          for which there is no such declaration
                        </li>
                      </ul>
                      If these rules produce no clear winner, the
                      source document is invalid.
                    </li>
                  </ul>
                </div>
                <p>
                  A side effect of this substitution is the generation of an annotation assertion that preserves the original shorthand relation:
                  <span style="white-space: pre;">AnnotationAssertion(<span class="name">T</span>(shorthand) <span class="name">T</span>(Formal-ID) "Original-ID"^^xsd:string)</span>
                </p>
                <p>
                  These rules are normative. An informative summary is provided in section <a href="#8.2.2">8.2.2</a>.
                </p>

<h4>Mapping ID Prefixes</h4>

<ul>
  <li>
    If the OBODoc header tag contains an idspace clause <b>idspace(IDPrefix URIPrefix)</b>, then the value of <span class="name">T<sup>prefix</sup></span>(IDPrefix) is URIPrefix
  </li>
  <li>
    If the OBODoc header tag contains no matching idspace clause then the value of of <span class="name">T<sup>prefix</sup></span>(prefix) is 'http://purl.obolibrary.org/obo/'
  </li>
</ul>

<h4>Examples (informative)</h4>




<h3><a name="5.10"></a>5.10 Post-Processing OWL DL</h3>

<table border="1" cellspacing="0">
<caption><a id="postProcessing" name="postProcessing" ></a>
  Post-processing OWL2 DL
</caption>  
<thead>
  <tr><th>Remove Axiom</th>
  <th>If Ontology Contains</th>
    <th>Replace With</th></tr>
</thead>

<tbody>
  <tr>
    <td>
      ObjectExactCardinality(n P CE), n&gt;0
    </td>
    <td>
      TransitiveProperty(P)
    </td>
    <td>
      SomeValuesFrom(P CE)
    </td>
  </tr>
  <tr>
    <td>
      ObjectMinCardinality(n P CE), n&gt;0
    </td>
    <td>
      TransitiveProperty(P)
    </td>
    <td>
      SomeValuesFrom(P CE)
    </td>
  </tr>
  <tr>
    <td>
      ObjectMaxCardinality(n P CE)
    </td>
    <td>
      TransitiveProperty(P)
    </td>
    <td>
      -
    </td>
  </tr>
</tbody>
</table>

<p>
</p>

<h2><a name="5.11"></a>5.11 OWL to OBO</h2>

                <p>
                  Not every OWL ontology can be converted to OBO
                  Format. This may be because the OWL ontology uses
                  constructs that cannot be expressed directly in OBO,
                  or it may be because the corresponding OBO ontology
                  would violate OBO document structure constraints
                  (see section <a href="#4.4">4.4</a>).
                </p>
                <div>
                  <ul>
                    <li>If a Frame has more than one name after translation, an arbitrary one is selected, and a warning is fired</li>
                    <li>If a Frame has more than one comment after translation, the set of distinct comments for that frame is concatenated using the \n escape character</li>
                    <li>If an expression cannot be translated, it is dropped OR converted into an OWL Macro, see below</li>
                  </ul>
                </div>
<p>
  Note that all classes in an OBO Document are assumed to
  satisfiable. Any axioms refering to owl:Thing should be removed. An
  ontology with unsatisfiable classes is not translateable.
</p>


<h2><a name="6"></a>6 OBO Sublanguages</h2>

<p>
We introduce the concept of an OBO sublanguage. This is similar to an
                  OWL profile, but the goal is not to provide
                  sublanguages for reasoning purposes. Sublanguages
                  are intended for legacy applications that make
                  unwarranted assumptions about OBO format. The
                  assumptions do not hold about OBO documents as a
                  whole, but they may hold for particular
                  sublanguages.
</p>
<p>
At this time, only one sublanguage is specified - OBO basic. To
                  define this sublanguage, we first introduce a number
                  of sublanguage characteristics.
</p>

<h3><a name="6.1"></a>6.1 OBO Sublanguage Characteristics</h3>
<h3><a name="6.1.1"></a>6.1.1 DAG</h3>

<div>
An ontology is a DAG if the graph formed by its logical relationships
does not contain any cycles. Formally: let V be a set of nodes
corresponding to all Term frames in the ontology (OWL Classes). Let E
be a set of edge pairs A,B where both A and B are in V and either
<ul>
<li> A is_a B (i.e. SubClassOf(A B)) or</li>
<li> relationship: R B (i.e. SubClassOf(A ObjectSomeValuesFrom(R B)))</li>
</ul>
Note that the relationship type (i.e. ObjectProperty) is irrelevant here.                    
</div>
<p>
The ontology is a DAG if the transitive closure of E contains no cycles.
</p>

<p>
Note that any ontology that contains an EquivalenceAxiom between two
named classes is not a DAG, as an equivalence axiom is the same as a
reciprocal SubClassOf pair.
</p>

<b>Rationale:</b> the original Gene Ontology paper described the GO as
                a DAG. Consequently a large body of legacy software
                exists that assumes the ontology is a DAG. This
                software breaks or loops indefinitely if there is a
                cycle involving any set of is_a or relationship tags

<h3><a name="6.1.2"></a>6.1.2 Dangling Clauses</h3>

<p>
A dangling clause is an is_a, relationship, intersection_of, union_of,
transitive_over, equivalent_to_chain, holds_over_chain, disjoint_from,
domain or range clause in which any of the values is not declared in
the ontology document. Here declaration means there exists a frame
with the corresponding identifier.
</p>

<div>
We can define 3 sub-characteristics:
<ul>
 <li>
                      Dangling Classes
 </li>
 <li>
                      Dangling Relations
 </li>
 <li>
                      Dangling Individuals
 </li>
</ul>

</div>

<h3><a name="6.1.3"></a>6.1.3 Unidirectional</h3>

<!-- TODO: define in terms of unfoldability -->

<p>
An ontology is said to be multidirectional if it contains a class
assertion involving property P and a class assertion involving
property P', where P is the inverse of P'. An ontology is
unidirectional if it is not multidirectional.
</p>
<p>
For example, if an ontology contains statements using both part_of and
has_part properties, then the ontology is not unidirectional.
</p>
<p>
Note that multidirectional ontologies with frequently be non-DAGs,
according to the definition above, but this need not hold.
</p>

<h3><a name="6.1.4"></a>6.1.4 Fully asserted</h3>

<!-- TODO: simplify -->

<p>
An ontology is fully asserted if the graph formed by is_a and
relationship clauses (i.e. SubClassOf axioms) is not lacking any
relationships that can be inferred from the full set of axioms in the
ontology (e.g. EquivalentClasses axioms, or DisjointFrom Axioms).
</p>

<p>
More formally: O is an ontology, and O' is an ontology consisting only
of is_a and relationship tags asserted in O. O is fully-asserted if
there is no SubClassOf axiom that is entailed by O but not by O'.
</p>

<p>
If an ontology is fully asserted, then basic applications can use
solely is_a and relationship tags for graph traversal. intersection_of
tags can be removed with no loss of functionality in these basic
applications. Ontology publishers may wish to keep two versions - an
edit version, in which only minimal assertions are made, and a release
version, which is fully asserted, with the assertions added by a
reasoner.
</p>

<h3><a name="6.1.5"></a>6.1.5 Fully Labeled</h3>

<p>
An OBO ontology is fully labeled if and only if every frame has a "name" tag.
</p>

<h3><a name="6.1.6"></a>6.1.6 Uniquely Labeled</h3>

<p>
An OBO ontology is uniquely labeled if and only if there are no two
entities share the same name, including both the source ontology and
its full import closure.
</p>

<h3><a name="6.1.7"></a>6.1.7 Locally Uniquely Labeled</h3>

<p>
An OBO ontology is locally uniquely labeled if and only if there are
no two entities share the same name, considering only the source
ontology but not its import closure.
</p>

<h3><a name="6.1.8"></a>6.1.8 No equivalence axioms</h3>

<p>
An OBO document has no equivalence axioms if the OWL translation 
contains no EquivalentClasses axioms.
</p>

<h3><a name="6.1.9"></a>6.1.9 No obo-namespace crossing</h3>

<p>
An OBO document has no obo-namespace crossing if there are no is_a,
                  relationship, intersection_of or union_of clauses
                  that span two obo-namespaces. Note that an
                  obo-namespace is not the same as an OWL
                  namespace. It is specified by the namespace tag in
                  OBOF. An example is found in the GO, which is a
                  single ontology partitioned into 3
                  obo-namespaces. Some legacy software assumes no
                  namespace crossings, so the basic version of the GO
                  released by the GOC has no obo-namespace crossings.
</p>

<h3><a name="6.1.10"></a>6.1.10 Singly labeled edges</h3>

<p>
An OBO document has singly labeled edges if there are no classes A and
B such that there are two asserted relationships connecting A and B
</p>

<h3><a name="6.1.11"></a>6.1.11 No disjointness axioms</h3>

<p>
An ontology contains no disjointness axioms if it has no
DisjointClasses, DisjointUnion or DisjointObjectProperties axioms. On
the obo-format level this manifests as no "disjoint_from:" tags.
</p>

<p>
Rationale: Some OBO-format based tools do not handle disjointness axioms well
</p>

<h3><a name="6.1.11"></a>6.1.11 No qualifier lists</h3>

<p>
An OBO document has no qualifier blocks if the OBO syntax
concretization has no '{...}'s appearing in the qualifier list
position. I.e. the following grammar rule (from <a href="#3.2">section 3.2</a>):
</p>

<p>
<span class="nonterminal">QualifierBlock</span> ::= '{' <span class="nonterminal">QualifierList</span> '}' 
</p>

<p>
Can be replaced by:
<span class="nonterminal">QualifierBlock</span> ::= 
</p>

<p>
On the OWL document level, this is equivalent to an OWL document with
no Axiom Annotations <i>except</i> those axiom annotations required to
support OBO metadata constructs, such as definitions and synonyms.
</p>

<h3><a name="6.1.12"></a>6.1.12 No owl-axioms header</h3>

<p>
See <a href="#5.0.4">5.0.4</a>. An ontology has no owl-axioms header
if the header does not contain an "owl-axioms:" tag.
</p>

<h3><a name="6.1.13"></a>6.1.13 No imports</h3>

<p>
An ontology has no imports if the header does not contain any
"import:" tags
</p>


<h3><a name="6.2"></a>6.2 OBO Basic</h3>

<p>
OBO Basic is a sub-language of OBO format (i.e. every OBO Basic
document is automatically an OBO Document). It allows basic
applications to continue to make certain simplifying assumptions; many
of these simplifying assumptions were based on the initial version of
the Gene Ontology, and have become enshrined in many popular and
useful tools such as term enrichment tools.
</p>
<p>
Examples of such assumptions include: traversing the ontology graph
ignoring relationship types using a naive algorithm will not lead to
cycles (i.e. the ontology is a DAG); every referenced term is declared
in the ontology (i.e. there are no dangling clauses).
</p>
<div>
An ontology is OBO Basic if and only if it has the following characteristics:                  

                <ul>
                  <li>
                    DAG
                  </li>
                  <li>
                    Unidirectional
                  </li>
                  <li>
                    No Dangling Clauses
                  </li>
                  <li>
                    Fully Asserted
                  </li>
                  <li>
                    Fully Labeled
                  </li>
                  <li>
                    No equivalence axioms
                  </li>
                  <li>
                    Singly labeled edges
                  </li>
                  <li>
                    No qualifier lists
                  </li>
                  <li>
                    No disjointness axioms
                  </li>
                  <li>
                    No owl-axioms header
                  </li>
                  <li>
                    No imports
                  </li>
                </ul>
</div>


<h3><a name="6.3"></a>6.3 Conversion to sub-languages</h3>

<p>
In general translation to a sublanguage is lossy, but this is not
always the case. An OBO-Dangling-Clause ontology can be converted to a
semantically equivalent ontology with declarations for all undeclared
referenced frames (note that this typically happens as a side-effect
of obo-to-ow-to-obo roundtripping). Note that this means the target
ontology is no longer OBO-Fully-Labeled.
</p>


<h2><a name="7"></a>7 OWL Macros</h2>

<p>
  This section cab be considered an independent component that can be
  used for any OWL ontology. It may eventually form its own separate
  specification. It should be seen as informative for now.
</p>

<p>
  Macros are specified as the range of an AnnotationAssertion on a
  property. They take the form of a string encoding either an OWL
  axiom or OWL expression in an extension Manchester Syntax format
  [<cite><a href="#ref-bcp-47" title="">BCP 47</a></cite>]. This
  extension allows the use of variable markers (written <i>?X</i>
  or <i>?Y</i>) in place of classes or class expressions. When these
  variable markers are replaced by actual OWL classes or expressions
  (again serialized in Manchester Syntax) as part of a Template
  Substitution Operation <span class="name">Subst</span>, the
  resulting string must be valid Manchester Syntax.
</p>

<p>
  Examples are provided in [<cite><a href="#owl-macros" title="">OWL Macros</a></cite>].
</p>

<h2><a name="7.1"></a>7.1 Macro Annotation Assertions</h2>

<p>
Two IRIs are used to specify the two different types of macro expansion.
</p>

<table border="1" cellspacing="0">
<caption><a id="translation-q" name="translationQ" ></a>
	Macros</caption>

<thead>
<tr>
  <th>Macro</th>
  <th>Annotation Property</th>
  <th>Variables</th>
  <th>Generates</th>
</thead>

<tbody>

<tr>
<td valign="top" style="white-space: pre;">expand_expression_to</td>
<td valign="top" style="white-space: pre;">obo:IAO_0000424</td>
<td valign="top" style="white-space: pre;"><i>?Y</i></td>
<td valign="top" style="white-space: pre;">OWL2 Class Expression</td>
</tr>

<tr>
<td valign="top" style="white-space: pre;">expand_assertion_to</td>
<td valign="top" style="white-space: pre;">obo:IAO_0000425</td>
<td valign="top" style="white-space: pre;"><i>?X</i>,<i>?Y</i></td>
<td valign="top" style="white-space: pre;">OWL2 Axiom</td>
</tr>

</tbody>
</table>

<h2><a name="7.2"></a>7.2 Macro Expansion and Replacement</h2>

<p>
</p>

<table border="1" cellspacing="0">
<caption><a id="postProcessing" name="postProcessing" ></a>
  OWL Macro Expansion Rules
</caption>  
<thead>
  <tr><th>Expression or Axiom to be Replaced</th>
  <th>Condition</th>
    <th>Replace With</th></tr>
</thead>

<tbody>

<tr>
    <td>
      ObjectSomeValuesFrom(P CE)
    </td>
    <td>
      AnnotationAssertion(expand_expression_to P Template)
    </td>
    <td>
      <span class="name">Subst</span>(Template)[CE]
    </td>
</tr>

<tr>
    <td>
      ObjectAllValuesFrom(P CE)
    </td>
    <td>
      AnnotationAssertion(expand_expression_to P Template)
    </td>
    <td>
      <span class="name">Subst</span>(Template)[CE]
    </td>
</tr>

<tr>
    <td>
      AnnotationAssertion(P A B)
    </td>
    <td>
      AnnotationAssertion(expand_assertion_to P Template)
    </td>
    <td>
      <span class="name">Subst</span>(Template)[A,B]
    </td>
</tr>

</tbody>
</table>

<p>

</p>

<h2><a name="7.3"></a>7.3 Macro-based GCI Enhancement</h2>

                <p>
                  An alternative strategy is to use the occurrence of
                  macro relations to add General Class Inclusion (GCI)
                  axioms to the ontology
                </p>

<table border="1" cellspacing="0">
<caption><a id="postProcessing" name="postProcessing" ></a>
  GCI Addition Rules
</caption>  
<thead>
  <tr><th>Matched Expression or Axiom</th>
  <th>Condition</th>
    <th>GCI to Add</th></tr>
</thead>

<tbody>

<tr>
    <td>
      ObjectSomeValuesFrom(P CE)
    </td>
    <td>
      AnnotationAssertion(expand_expression_to P Template)
    </td>
    <td>
      EquivalentClasses(ObjectSomeValuesFrom(P CE) <span class="name">Subst</span>(Template)[CE])
    </td>
</tr>

<tr>
    <td>
      ObjectAllValuesFrom(P CE)
    </td>
    <td>
      AnnotationAssertion(expand_expression_to P Template)
    </td>
    <td>
      EquivalentClasses(ObjectSomeValuesFrom(P CE) <span class="name">Subst</span>(Template)[CE])
    </td>
</tr>

<tr>
    <td>
      AnnotationAssertion(P A B)
    </td>
    <td>
      AnnotationAssertion(expand_assertion_to P Template)
    </td>
    <td>
      use previous table
    </td>
</tr>

</tbody>
</table>

<h2><a name="7.4"></a>7.4 Placement of Generated Axioms in External Ontologies</h2>

<p>
  The behavior of a macro-expansion engine regarding whether to
  replace on add new GCIs is implementation-dependent, as this
  decision of whether to place generated axioms in a new ontology or
  the source ontology.
                </p>
                <div>
                  The <i>recommended behavior</i> is:
                  <ul>
                    <li>
                      The 2nd (GCI) table should be used.
                    </li>
                    <li>
                      The source ontology should be left unmodified.
                    </li>
                    <li>
                      GCIs generated from expand_expression_to
                      templates and new axioms generated by
                      expand_assertion_to should be placed in a
                      separate ontology or ontologies. The recommended
                      IRI of this ontology will be specified in the
                      id-policy document.
                    </li>
                  </ul>
</div>

<h2><a name="8"></a>8 Recommendations (informative)</h2>

<h3><a name="8.1"></a>8.1 Generating OBO Format Documents</h3>
                <p>
                  These guidelines are provided to minimize character-level differences between OBO Documents generated by different procedures.
                </p>
<h3><a name="8.1.1"></a>8.1.1 Whitespace</h3>
                <ul>
                  <li>
                    For every tag-value line, the initial ':' character should be followed by a space character (' ').
                  </li>
                  <li>
                    Every successive pair of frames should be seperated by a single empty line.
                  </li>
                  <li>
                    The initial frame should be separated from the header block by a single empty line.
                  </li>
                  <li>
                    Each document should have an ontology declaration in the header.
                  </li>
                  <li>
                    The ontology id should be all lower case, and
                    should match the primary ID space used in the
                    ontology (for example: 'go', 'cl').
                  </li>
                </ul>
<h3><a name="8.1.2"></a>8.1.2 File Comments</h3>
                <ul>
                  <li>
                    If a frame references an identifier, and that
                    identifier is opaque (i.e. it conforms to the
                    Canonical-Prefixed-ID production rule), then the
                    generator should add commments, adding a label for
                    every opaque identifier. For example:
                    <pre>
relationship: part_of ABC:1234567 ! hand
relationship: R:9999999 ABC:1234567 ! part_of hand
                    </pre>
                  </li>
                  <li>
                    All file comments should be preceded by a tag-value pair, and there should be exactly one space character on either side of the '!' character
                  </li>
                </ul>
<h3><a name="8.1.3"></a>8.1.3 Tag Ordering</h3>
                <p>
                  See
                  the <a href="http://www.geneontology.org/GO.format.obo-1_4.shtml#S.3.5">serializer
                  conventions</a> in the OBO format guide
                </p>
<h3><a name="8.1.1"></a>8.1.4 Escaping characters</h3>
<p>
  When writing solitary 'xref' tags, the comma character should be escaped.
</p>
<h3><a name="8.2"></a>8.2 Identifiers</h3>
<h3><a name="8.2.1"></a>8.2.1 Class (Term) Identifiers</h3>
                <p>
                  All class identifiers should
                  follow <a href="http://www.obofoundry.org/id-policy.shtml">OBO
                  Foundry ID Policy</a>. i.e. all ID-spaces must be
                  registered and approved (email obo-admin at
                  obofoundry.org to register), and the entire ID
                  should consist of the ID space followed by ':'
                  followed by a zero-padded numeric local identifer (a
                  minimum of 7 digits is recommended).
                </p>
<h3><a name="8.2.2"></a>8.2.2 Relation (Property) Identifiers</h3>
                <p>
                  Relation identifiers should in follow the same
                  guidelines as class identifiers. Note however that
                  the use of symbolic identifiers such as 'part_of' is
                  common in almost all OBO format ontologies, and has
                  a precedent stretching back over ten years. A large
                  body of software now expects symbolic identifiers
                  for relations, and ontology maintainers are
                  understandably reluctant to change these to numeric
                  identifiers.
                </p>
                <p>
                  This specification provides a means of using numeric
                  identifiers globally whilst retaining symbolic
                  identifiers within the context of a single
                  file. Refer to section <a href="#5.9.3">5.9.3</a> for
                  details.
                </p>
                <ul>
                  <li>
                    Every symbolic relation identifier
                    (e.g. 'part_of') should have an xref tag to a formal relation identifier. E.g.
<pre>
[Typedef]
id: has_part
name: has_part
xref: BFO:0000051
</pre>
                  </li>
                  <li>
                    This xref should refer to either BFO
                    (<a href="http://purl.obolibrary.org/obo/bfo.owl">http://purl.obolibrary.org/obo/bfo.owl</a>)
                    or to RO
                    (<a href="http://purl.obolibrary.org/obo/ro.owl">http://purl.obolibrary.org/obo/ro.owl</a>). Xrefs
                    to the old RO can be provided for historic purposes, but are otherwise discouraged.
                  </li>
                  <li>
                    According to the rules in
                    section <a href="#5.9.3">5.9.3</a>, the symbolic
                    relation identifier can be used as a shorthand for
                    the formal relation identifier.
                  </li>
                  <li>
                    When roundtripping the OBO file, the symbolic identifiers should be preserved.
                  </li>
                </ul>
<h3><a name="8.3"></a>8.3 Individuals</h3>
<p>
Individuals are retained in OBOF1.4 for compatibility with previous
                  versions of the format. OBOF is not a good choice
                  for modeling using individuals. Ontologies should
                  not contain individuals, and OBOF is not a good
                  choice for knowledge bases that use individuals.
</p>

<h3><a name="8.4"></a>8.4 Imports</h3>
<p>
When translating between OBO and OWL, owl:imports in an OWL ontology
hader are translated directly to an 'import:' header in an OBO
document. The contents of the field MUST not be modified during
translation (normative).
</p>
<p>
As a consequence of this, URIs that resolve to OWL documents must be
used if the resulting OWL is to be consumed by an OWL parser. Here is
an example of a valid OBO document header (taken from the GO editors file):
<pre>
import: http://purl.obolibrary.org/obo/go/extensions/chebi_import.owl
import: http://purl.obolibrary.org/obo/go/extensions/cl_import.owl
import: http://purl.obolibrary.org/obo/go/extensions/po_import.owl
ontology: go
</pre>
<p>
The onus is on the OBO parser to be able to handle these OWL documents
- either by translating them or mapping the URIs to OBO documents that
have been translated in advance

</p>
<p>

Most OBO tools are not equipped to do this, so imports should be used
with caution in OBO format. We strongly recommend using OWL whenever
imports are required, and providing OBO versions with the imports
closure merged. Note that this can be done with owltools:

<pre>
owltools myont.owl --merge-imports-closure -o -f obo myont.obo
</pre>

Note that all versions of OBO-Edit prior to 2.3b7 should be considered
broken for imports. 2.3b7 introduces the use of oboedit.catalog files
which allow OWL URIs to be used in import directives. These are not
part of the OBOFormat spec, but details are included here for completeness

<h4><a name="8.4.1"></a>8.4.1 OBO-Edit catalog files</h4>

An OBO-Edit catalog file contains newline separated entries mapping an
OWL URI to a local file containing an OBO document, such that the OBO
document can be substituted for the OWL URI with no loss of
content. Each mapping uses a single line, and is separated by
whitespace.

The file can optionally include lines starting with "#"
which are ignored and for human consumption.

An example:

<pre>
# GO imports mappings
http://purl.obolibrary.org/obo/go/extensions/chebi_import.owl ../extensions/chebi_import.obo
http://purl.obolibrary.org/obo/go/extensions/cl_import.owl ../extensions/cl_import.obo
http://purl.obolibrary.org/obo/go/extensions/po_import.owl ../extensions/po_import.obo
</pre>

<p>
When an OBO parser encounters an import directive it may choose to
load the document obtained via mapping the URI on the left with the
OBO document found in the file on the right.

The oboedit.catalog file should live in the same directory as the
source OBO ontology.

</p>

<a name="References"></a><h2> <span class="mw-headline">9  References </span></h2>
<a name="Normative_References"></a><h3> <span class="mw-headline">9.1  Normative References </span></h3>
<dl><dt> <span id="ref-bcp-47">[BCP 47]</span>
</dt><dd> <cite><a class="external text" href="http://www.rfc-editor.org/rfc/bcp/bcp47.txt" title="http://www.rfc-editor.org/rfc/bcp/bcp47.txt">BCP 47 - Tags for Identifying Languages</a></cite>. A. Phillips and M. Davis, eds.  IETF, September 2006. http://www.rfc-editor.org/rfc/bcp/bcp47.txt
</dd><dt> <span id="ref-owl-2-specification">[OWL 2 Specification]</span>
</dt><dd><span><cite><a href="http://www.w3.org/TR/2009/REC-owl2-syntax-20091027/">OWL 2 Web Ontology Language: <span>Structural Specification and Functional-Style Syntax</span></a></cite> Boris Motik, Peter F. Patel-Schneider, Bijan Parsia, eds. W3C Recommendation, 27 October 2009, <a href="http://www.w3.org/TR/2009/REC-owl2-syntax-20091027/">http://www.w3.org/TR/2009/REC-owl2-syntax-20091027/</a>.  Latest version available at <a href="http://www.w3.org/TR/owl2-syntax/">http://www.w3.org/TR/owl2-syntax/</a>.</span></dd><dt> <span id="ref-rdf-testcases">[RDF Test Cases]</span>
</dt><dd> <cite><a class="external text" href="http://www.w3.org/TR/2004/REC-rdf-testcases-20040210/" title="http://www.w3.org/TR/2004/REC-rdf-testcases-20040210/">RDF Test Cases</a></cite>. Jan Grant and Dave Beckett, eds.  W3C Recommendation, 10 February 2004, http://www.w3.org/TR/2004/REC-rdf-testcases-20040210/.  Latest version available as http://www.w3.org/TR/rdf-testcases/.
</dd><dt> <span id="ref-rfc-3629">[<a class="external" href="http://tools.ietf.org/html/rfc3629" title="http://tools.ietf.org/html/rfc3629">RFC 3629</a>]</span>
</dt><dd> <cite><a class="external text" href="http://www.ietf.org/rfc/rfc3629.txt" title="http://www.ietf.org/rfc/rfc3629.txt">RFC 3629: UTF-8, a transformation format of ISO 10646</a></cite>.  F. Yergeau.  IETF, November 2003, http://www.ietf.org/rfc/rfc3629.txt
</dd><dt> <span id="ref-rfc-3987">[RFC 3987]</span>
</dt><dd> <cite><a class="external text" href="http://www.ietf.org/rfc/rfc3987.txt" title="http://www.ietf.org/rfc/rfc3987.txt">RFC 3987: Internationalized Resource Identifiers (IRIs)</a></cite>. M. Duerst and M. Suignard. IETF, January 2005, http://www.ietf.org/rfc/rfc3987.txt
</dd><dt> <span id="ref-sparql">[SPARQL]</span>
</dt><dd> <cite><a class="external text" href="http://www.w3.org/TR/2008/REC-rdf-sparql-query-20080115/" title="http://www.w3.org/TR/2008/REC-rdf-sparql-query-20080115/">SPARQL Query Language for RDF</a></cite>. Eric Prud'hommeaux and Andy Seaborne, eds. W3C Recommendation, 15 January 2008, http://www.w3.org/TR/2008/REC-rdf-sparql-query-20080115/.  Latest version available as http://www.w3.org/TR/rdf-sparql-query/.
</dd><dt> <span id="ref-unicode">[UNICODE]</span>
</dt><dd> <cite><a class="external text" href="http://www.unicode.org/unicode/standard/" title="http://www.unicode.org/unicode/standard/">The Unicode Standard</a></cite>. The Unicode Consortium, Version 5.1.0, ISBN 0-321-48091-0, as updated from time to time by the publication of new versions. (See <a class="external free" href="http://www.unicode.org/unicode/standard/versions/" title="http://www.unicode.org/unicode/standard/versions/">http://www.unicode.org/unicode/standard/versions/</a> for the latest version and additional information on versions of the standard and of the Unicode Character Database).
</dd></dl>

<a name="Non-normative_References"></a><h3> <span class="mw-headline">8.2  Non-normative References </span></h3>

<dl>

  <dt> <span id="obo-owl-horrocks">[OBO-OWL Horrocks]</span>
  </dt><dd> <cite><a class="external text" href="http://www.cs.man.ac.uk/~horrocks/obo/" title="http://www.cs.man.ac.uk/~horrocks/obo/">OBO Flat File Format Syntax and Semantics
and Mapping to OWL Web Ontology Language</a></cite>. Ian Horrocks, March 2007. http://www.cs.man.ac.uk/~horrocks/obo/
  </dd>

  <dt> <span id="obo-owl-golbreich">[OBO-OWL Golbreich]</span>
  </dt><dd> <cite><a class="external text" href="http://en.scientificcommons.org/43278282" title="http://en.scientificcommons.org/43278282">The OBO to OWL mapping, GO to OWL 1.1</a></cite>. Christine Golbreich, Ian Horrocks, 2007. http://en.scientificcommons.org/43278282
  </dd>

  <dt> <span id="obo-owl-ncbo">[OBO-OWL NCBO]</span>
  </dt><dd> <cite><a class="external text" href="http://www.bioontology.org/wiki/index.php/OboInOwl:Main_Page" title="http://www.bioontology.org/wiki/index.php/OboInOwl:Main_Page">NCBO Mapping OBO to OWL</a></cite>. Dilvan Moreira, John Day-Richter, Chris Mungall, Nigam H. Shah 2007, http://www.bioontology.org/wiki/index.php/OboInOwl:Main_Page
  </dd>

  <dt> <span id="obolog">[Obolog]</span>
  </dt><dd> <cite><a class="external text" href="http://oboedit.org/obolog/spec/obolog-spec.pdf" title="http://oboedit.org/obolog/spec/obolog-spec.pdf">Obolog specification</a></cite>. Chris Mungall. Aug 2008, http://oboedit.org/obolog/spec/obolog-spec.pdf
  </dd>

  <dt> <span id="obo-owl">[OBO-OWL WG]</span>
  </dt><dd> <cite><a class="external text" href="http://www.obofoundry.org/wiki/index.php/Working_Group_on_OBO-OWL_interconversion" title="http://www.obofoundry.org/wiki/index.php/Working_Group_on_OBO-OWL_interconversion">Working Group on OBO-OWL interconversion</a></cite>. http://www.obofoundry.org/wiki/index.php/Working_Group_on_OBO-OWL_interconversion
  </dd>

  <dt> <span id="obo-owl-jbms">[OBO OWL JBMS]</span>
  </dt><dd> <cite><a class="external text" href="http://userweb.cs.utexas.edu/~hamid/oboowl.html" title="http://userweb.cs.utexas.edu/~hamid/oboowl.html">Mapping between the OBO and OWL ontology languages.</a></cite>. Syed Hamid Tirmizi, Stuart Aitken, Dilvan Moreira, Chris Mungall, Juan Sequeda, Nigam H. Shah and Daniel P. Miranker. In: Proceedings of Semantic Web Applications and Tools for Life Sciences Workshop, Amsterdam, The Netherlands,
  </dd>

  <dt> <span id="obo-owl-swat4ls">[OBO OWL SWAT4LS]</span>
  </dt><dd> <cite><a class="external text" href="http://userweb.cs.utexas.edu/~hamid/oboowl.html" title="http://userweb.cs.utexas.edu/~hamid/oboowl.html">OBO &amp; OWL: Roundtrip Ontology Transformations.</a></cite>. Syed Hamid Tirmizi, Stuart Aitken, Dilvan Moreira, Chris Mungall, Juan Sequeda, Nigam H. Shah and Daniel P. Miranker. In: Proceedings of Semantic Web Applications and Tools for Life Sciences Workshop, Amsterdam, The Netherlands,
  </dd>

  <dt> <span id="obo-1-4">[OBO Format 1.4 Guide]</span>
  </dt><dd> <cite><a class="external text" href="http://www.geneontology.org/GO.format.obo-1_4.shtml" title="http://www.geneontology.org/GO.format.obo-1_4.shtml">OBO Format 1.4 Guide</a></cite>. Chris Mungall and John-Day Richter, August 2010, http://www.geneontology.org/GO.format.obo-1_4.shtml
  </dd>

  <dt> <span id="ref-manchester-owl-dl">[Manchester OWL DL Syntax]</span>
  </dt><dd> <cite><a class="external text" href="http://www.w3.org/TR/owl2-manchester-syntax/" title="http://www.w3.org/TR/owl2-manchester-syntax/">OWL 2 Web Ontology Language
Manchester Syntax</a></cite>. Matthew Horridge, Peter F. Patel-Schneider, October 2009, http://www.w3.org/TR/owl2-manchester-syntax/
  </dd>

  <dt> <span id="owl-macros">[OWL Macros]</span>
  </dt><dd> <cite><a class="external text" href="http://precedings.nature.com/documents/5292/version/1" title="http://precedings.nature.com/documents/5292/version/1">OWL 2 Macros</a></cite>. Chris Mungall, David Osumi-Sutherland, Alan Ruttenberg, June 2010, http://precedings.nature.com/documents/5292/version/1
  </dd>

  <dt> <span id="owldef">[OWLDEF]</span>
  </dt><dd> <cite><a class="external text" href="http://leechuck.de/pub/or-sig.pdf" title="http://leechuck.de/pub/or-sig.pdf">OWLDEF: Integrating OBO and OWL</a></cite>. R Hoehndorf, A Oellrich, M Dumontier, J Kelso, H Herre, http://leechuck.de/pub/or-sig.pdf
  </dd>

  <dt> <span id="rel-patterns">[Relational Patterns]</span>
  </dt><dd> <cite><a class="external text" href="http://www.webont.org/owled/2010/papers/owled2010_submission_3.pdf" title="http://www.webont.org/owled/2010/papers/owled2010_submission_3.pdf">Relational patterns in OWL and their application to OBO</a></cite>. R Hoehndorf, A Oellrich, M Dumontier, J Kelso, H Herre, "http://www.webont.org/owled/2010/papers/owled2010_submission_3.pdf
  </dd>

</dl>
</body>
</html>