# TFBS module for TFBS::DB::JASPAR # # Copyright Boris Lenhard # # You may distribute this module under the same terms as perl itself # # Maintainers: # Xiaobei Zhao - JASPAR6 (JASPAR 2014 experimental) # David Arenillas - JASPAR7 (JASPAR 2016) # # This is an update of, and follows from TFBS::DB::JASPAR6 which itself was # created / modified from JASPAR5 by Xiaobei Zhao. It was created # by copying the existing TFBS::DB::JASPAR6 and modifying it to reflect the # changes made to the 2016 update of the JASPAR database / webserver. # # As there seemed to be some discrepancy / confusion between the version # numbers with the JASPAR DB/webserver releases and the perl module, from # now on the latest (current) version of the perl module will simply be named # JASPAR.pm (without version number attached). Previous versions will have a # version number attached to them to indicate which DB/webserver they are # related to. # # JASPAR 5 and 6 are associated with JASPAR 2014 (also known as JASPAR 5.0). # It appears that the JASPAR6 perl module was an experimental update to # JASPAR5 and did not reflect a version change in the JASPAR DB/webserver. # # Change summary since JASPAR[5/6].pm: # - In the _store_matrix routine, the code was checking for a version number # by checking for a 'version' tag and NOT even checking to see if the version # number was already contained in the matrix ID!?!? Modified to first check # for a version encoded in the matrix ID and then only if it isn't, check for # a version tag. # - Fixed the way the 'acc' tag is stored. # - The DB schema was changed to allow multiple values for a given matrix in # some instances where previously only a single value was allowed. E.g. for # the MATRIX_ANNOTATION table, removed the unique key constrain on ID + TAG # so that multiple VALs with the same TAG can be stored for a given matrix. # As it was, multiple values were already stored as comma separated values # in a single row which is bad DB design (not normalized). The Code was thus # modified to reflect that these values may now be stored in multiple DB # rows. # - Related to this, previously it was assumed that TF class and family could # each only be a single value. However, for heterodimers it is quite possible # that the two TFs making up the dimer have a different class/family so now # the code assumes multiple could exist in the DB for these tags. As a result # the class/family may now be stored in a TFBS::Matrix object as either a # scalar string (for a single value) or as a listref (for multiple values). # This is consistent with the way other tag/values are stored although # it can certainly be argued that it is not good programming style to have # different return types for the same method! This behaviour should be # reconsidered in future versions of the TFBS modules. # - Similarly to values stored in MATRIX_ANNOTATION, the MATRIX_SPECIES table # has been modified to allow storage or multiple species for a matrix as # separate rows rather than as comma separated strings and related code # modified accordingly. # - Added some exception calls and/or more informative error messages in some # places they were missing. # - Fixed up (some of) the ugly code formatting, e.g. changed mixture of # leading tabs / space to be consistently spaces and changed indentation # levels to always be 4 spaces. # # Also see specific embedded comments tagged with DJA for more details # # POD =head1 NAME TFBS::DB::JASPAR - interface to MySQL relational database of pattern matrices. Currently status: experimental. =head1 SYNOPSIS =over 4 =item * creating a database object by connecting to the existing JASPAR6-type database my $db = TFBS::DB::JASPAR6->connect("dbi:mysql:JASPAR6:myhost", "myusername", "mypassword"); =item * retrieving a TFBS::Matrix::* object from the database # retrieving a PFM by ID my $pfm = $db->get_Matrix_by_ID('M0079','PFM'); #retrieving a PWM by name my $pwm = $db->get_Matrix_by_name('NF-kappaB', 'PWM'); =item * retrieving a set of matrices as a TFBS::MatrixSet object according to various criteria # retrieving a set of PWMs from a list of IDs: my @IDlist = ('M0019', 'M0045', 'M0073', 'M0101'); my $matrixset = $db->get_MatrixSet(-IDs => \@IDlist, -matrixtype => "PWM"); # retrieving a set of ICMs from a list of names: @namelist = ('p50', 'p53', 'HNF-1'. 'GATA-1', 'GATA-2', 'GATA-3'); my $matrixset = $db->get_MatrixSet(-names => \@namelist, -matrixtype => "ICM"); =item * creating a new JASPAR6-type database named MYJASPAR6: my $db = TFBS::DB::JASPAR4->create("dbi:mysql:MYJASPAR6:myhost", "myusername", "mypassword"); =item * storing a matrix in the database (currently only PFMs): #let $pfm is a TFBS::Matrix::PFM object $db->store_Matrix($pfm); =back =head1 DESCRIPTION TFBS::DB::JASPAR is a read/write database interface module that retrieves and stores TFBS::Matrix::* and TFBS::MatrixSet objects in a relational database. The interface is nearly identical to the JASPAR2 and JASPAR4 interface, while the underlying data model is different =head1 JASPAR6 DATA MODEL JASPAR6 is working name for a relational database model used for storing transcriptional factor pattern matrices in a MySQL database. It was initially designed (JASPAR2) to store matrices for the JASPAR database of high quality eukaryotic transcription factor specificity profiles by Albin Sandelin and Wyeth W. Wasserman. Besides the profile matrix itself, this data model stores profile ID (unique), name, structural class, basic taxonomic and bibliographic information as well as some additional, and custom, tags. Here goes a moore thorough description on tables and IDs ----------------------- ADVANCED --------------------------------- For the developers and the curious, here is the JASPAR6 data model: MISSING TEXT HEER ON HOW IT WORKS It is our best intention to hide the details of this data model, which we are using on a daily basis in our work, from most TFBS users. Most users should only know the methods to store the data and which tags are supported. ------------------------------------------------------------------------- =head1 FEEDBACK Please send bug reports and other comments to the author. =head1 AUTHOR - Boris Lenhard Boris Lenhard EBoris.Lenhard@cgb.ki.seE =head1 APPENDIX The rest of the documentation details each of the object methods. Internal methods are preceded with an underscore. =cut # The code begins HERE: package TFBS::DB::JASPAR7; use vars qw(@ISA $AUTOLOAD); # we need all three matrices due to the redundancy in JASPAR2 data model # which will hopefully be removed in JASPAR3 use TFBS::Matrix::PWM; use TFBS::Matrix::PFM; use TFBS::Matrix::ICM; use TFBS::TFFM; use TFBS::MatrixSet; use Bio::Root::Root; use DBI; # use TFBS::DB; # eventually use strict; @ISA = qw(TFBS::DB Bio::Root::Root); ######################################################################### # CONSTANTS ######################################################################### use constant DEFAULT_CONNECTSTRING => "dbi:mysql:JASPAR_DEMO"; # on localhost use constant DEFAULT_USER => ""; use constant DEFAULT_PASSWORD => ""; ######################################################################### # PUBLIC METHODS ######################################################################### =head2 new Title : new Usage : DEPRECATED - for backward compatibility only Use connect() or create() instead =cut sub new { _new(@_); } =head2 connect Title : connect Usage : my $db = TFBS::DB::JASPAR6->connect("dbi:mysql:DATABASENAME:HOSTNAME", "USERNAME", "PASSWORD"); Function: connects to the existing JASPAR6-type database and returns a database object that interfaces the database Returns : a TFBS::DB::JASPAR6 object Args : a standard database connection triplet ("dbi:mysql:DATABASENAME:HOSTNAME", "USERNAME", "PASSWORD") In place of DATABASENAME, HOSTNAME, USERNAME and PASSWORD, use the actual values. PASSWORD and USERNAME might be optional, depending on the user's access permissions for the database server. =cut sub connect { #DONE # a more intuitive syntax for the constructor my ($caller, @connection_args) = @_; $caller->new(-connect => \@connection_args); } =head2 dbh Title : dbh Usage : my $dbh = $db->dbh(); $dbh->do("UPDATE matrix_data SET name='ADD1' WHERE NAME='SREBP2'"); Function: returns the DBI database handle of the MySQL database interfaced by $db; THIS IS USED FOR WRITING NEW METHODS FOR DIRECT RELATIONAL DATABASE MANIPULATION - if you have write access AND do not know what you are doing, you can severely corrupt the data For documentation about database handle methods, see L Returns : the database (DBI) handle of the MySQL JASPAR2-type relational database associated with the TFBS::DB::JASPAR2 object Args : none =cut sub dbh { #DONE my ($self, $dbh) = @_; $self->{'dbh'} = $dbh if $dbh; return $self->{'dbh'}; } =head2 store_Matrix Title : store_Matrix Usage : $db->store_Matrix($matrixobject); Function: Stores the contents of a TFBS::Matrix::DB object in the database Returns : 0 on success; $@ contents on failure (this is too C-like and may change in future versions) Args : (PFM_object) A TFBS::Matrix::PFM, FBS::Matrix::PWM or FBS::Matrix::ICM object. PFM object are recommended to use, as they are eaily converted to other formats # might have to give version and collection here Comment : this is an experimental method that is not 100% bulletproof; use at your own risk =cut sub store_Matrix { #PROBABLY DONE # collection, version are taken from the corresponding tags. Warn if they are not there ; my ($self, @PFMs) = @_; my $err; foreach my $pfm (@PFMs) { eval { my $int_id = $self->_store_matrix($pfm) ; # needs to have collection and version $self->_store_matrix_data($pfm, $int_id); $self->_store_matrix_annotation($pfm, $int_id); $self->_store_matrix_species($pfm, $int_id); $self->_store_matrix_acc($pfm, $int_id); }; } return $@; } sub create { #done my ($caller, $connectstring, $user, $password) = @_; if ( $connectstring and $connectstring =~ /dbi:mysql:(\w+)(.*)/) { # connect to the server; my $dbh = DBI->connect("dbi:mysql:mysql" . $2, $user, $password) or die("Error connecting to the database"); # create database and open it $dbh->do("create database $1") or die("Error creating database."); $dbh->do("use $1"); # create tables _create_tables($dbh); $dbh->disconnect; # run "new" with new database return $caller->new(-connect => [$connectstring, $user, $password]); } else { die( "Missing or malformed connect string for " . "TFBS::DB::JASPAR2 connection."); } } =head2 get_Matrix_by_ID Title : get_Matrix_by_ID Usage : my $pfm = $db->get_Matrix_by_ID('M00034', 'PFM'); Function: fetches matrix data under the given ID from the database and returns a TFBS::Matrix::* object Returns : a TFBS::Matrix::* object; the exact type of the object depending on what form the matrix is stored in the database (PFM is default) Args : (Matrix_ID) Matrix_ID id is a string which refers to the stable JASPAR ID (usually something like "MA0001") with or without version numbers. "MA0001" will give the latest version on MA0001, while "MA0001.2" will give the second version, if existing. Warnings will be given for non-existing matrices. =cut sub get_Matrix_by_ID { #DONE. MAYBE :) my ($self, $q, $mt) = @_; # q is a stable ID with possible version number # jsp6 $mt = (uc($mt) or "PFM"); unless (defined $q) { $self->throw("No ID passed to get_Matrix_by_ID"); } my $ucmt = uc $mt; # separate stable ID and version number my ($base_ID, $version) = split(/\./, $q); $version = $self->_get_latest_version($base_ID) unless $version; # latest version per default # get internal ID - also a check for validity my $int_id = $self->_get_internal_id($base_ID, $version); # get matrix using internal ID my $m = $self->_get_Matrix_by_int_id($int_id, $ucmt); #warn ref($m); return ($m); } =head2 get_Matrix_by_name Title : get_Matrix_by_name Usage : my $pfm = $db->get_Matrix_by_name('HNF-1'); Function: fetches matrix data under the given name from the database and returns a TFBS::Matrix::* object Returns : a TFBS::Matrix::* object; the exact type of the object depending on what form the matrix object was stored in the database (default PFM)) Args : (Matrix_name) Warning : According to the current JASPAR6 data model, name is not necessarily a unique identifier. Also, names change over time. In the case where there are several matrices with the same name in the database, the function fetches the first one and prints a warning on STDERR. You've been warned. Some matrices have multiple versions. The function will return the latest version. For specific versions, use get_Matrix_by_ID($ID.$version) =cut sub get_Matrix_by_name { #DONE my ($self, $name, $mt) = @_; unless (defined $name) { $self->throw("No name passed to get_Matrix_by_name."); } # sanity check: are there many different stable IDs with same name? my $sth = $self->dbh->prepare( qq!SELECT distinct BASE_ID FROM MATRIX WHERE NAME="$name"! ); $sth->execute(); my (@stable_ids) = $sth->fetchrow_array(); my $L = scalar @stable_ids; $self->warn("There are $L distinct stable IDs with name '$name'") if scalar $L > 1; return $self->get_Matrix_by_ID($stable_ids[0], $mt); } =head2 get_MatrixSet Title : get_MatrixSet Usage : my $matrixset = $db->get_MatrixSet(%args); Function: fetches matrix data under for all matrices in the database matching criteria defined by the named arguments and returns a TFBS::MatrixSet object Returns : a TFBS::MatrixSet object Args : This method accepts named arguments, corresponding to arbitrary tags, and also some utility functions Note that this is different from JASPAR2 and to some extent JASPAR4. As any tag is supported for database storage, any tag can be used for information retrieval. Additionally, arguments as 'name','class','collection' can be used (even though they are not tags. Per default, only the last version of the matrix is given. The only way to get older matrices out of this to use an array of IDs with actual versions like MA0001.1, or set the argyment -all_versions=>1, in which case you get all versions for each stable ID Examples include: Fundamental matrix features -all # gives absolutely all matrix entry, regardless of versin and collection. Only useful for backup situations and sanity checks. Takes precedence over everything else -ID # a reference to an array of stable IDs (strings), with or without version, as above. tyically something like "MA0001.2" . Takes precedence over everything salve -all -name # a reference to an array of # transcription factor names (string). Will only take latest version. NOT a preferred way to access since names change over time -collection # a string corresponding to a JASPAR collection. Per default CORE -all_versions # gives all matrix versions that fit with rest of criteria, including obsolete ones.Is off per default. # Typical usage is in combiation with a stable IDs withou versions to get all versinos of a particular matrix Typical tag queries: These can be either a string or a reference to an array of strings. If it is an arrau it will be interpreted as as an "or"s statement -class # a reference to an array of # structural class names (strings) -species # a reference to an array of # NCBI Taxonomy IDs (integers) -taxgroup # a reference to an array of # higher taxonomic categories (string) Computed features of the matrices -min_ic # float, minimum total information content # of the matrix. -matrixtype #string describing type of matrix to retrieve. If left out, the format will revert to the database format, which is PFM. The arguments that expect list references are used in database query formulation: elements within lists are combined with 'OR' operators, and the lists of different types with 'AND'. For example, my $matrixset = $db->(-class => ['TRP_CLUSTER', 'FORKHEAD'], -species => ['Homo sapiens', 'Mus musculus'], ); gives a set of TFBS::Matrix::PFM objects (given that the matrix models are stored as such) whose (structural clas is 'TRP_CLUSTER' OR'FORKHEAD') AND (the species they are derived from is 'Homo sapiens'OR 'Mus musculus'). As above, unless IDs with version numbers are used, only one matrix per stable ID wil be returned: the matrix with the highest version number The -min_ic filter is applied after the query in the sense that the matrices profiles with total information content less than specified are not included in the set. =cut # jsp6 sub get_MatrixSet { # IC conetent and matrix stuff is not there yet, rest should work my ($self, %args) = @_; #jsp6 $args{'-collection'} = 'CORE' unless $args{'-collection'}; $args{'-all_versions'} = 0 unless $args{'-all_versions'}; my @IDlist = @{$self->_get_IDlist_by_query(%args)} ; # the IDlist here are INTERNAL ids my $type; my $matrixset = TFBS::MatrixSet->new(); foreach my $int_id (@IDlist) { my $matrix = $self->_get_Matrix_by_int_id($int_id); if (defined $args{'-min_ic'}) { # we assume the matrix IS a PFM, o something in normal space at # least unless it explicitly says otherwise in tag=matrixtype # if so warn and do not use IC content # this is not foolproof in any way # # Fixed up logic to actually check $matrix->isa(TFBS::Matrix::ICM) # before checking the matrixtype tag. Also check that matrixtype # tag is defined before comparison to prevent annoying "Use of # uninitialized value in string eq" messages from perl. # DJA 2012/05/11 # if ($matrix->isa("TFBS::Matrix::ICM") || (defined $matrix->{tags}{matrixtype} && $matrix->{tags}{matrixtype} eq "ICM") ) { next if ($matrix->total_ic() < $args{'-min_ic'}); } elsif ($matrix->isa("TFBS::Matrix::PFM")) { next if ($matrix->to_ICM->total_ic() < $args{'-min_ic'}); } else { warn "Warning: you are assessing information content on matrices that are not in PFM or ICM format. Skipping this criteria"; next; } } # length if (defined $args{'-length'}) { next if ($matrix->length() < $args{'-length'}); } # number of sites within # since column sums MIGHT be slightly different we take the integer of the mean of the columns # or really int( sum of matrix/#columns) if (defined $args{'-sites'}) { my $sum = 0; foreach (1 .. $matrix->length) { $sum += $matrix->column_sum(); } $sum = int($sum / $matrix->length); #warn $matrix->ID, " $sum is $sum"; next if ($sum < $args{'-sites'}); } #ugly code: think about this a bit. if ($args{'-matrixtype'} && $matrix->isa("TFBS::Matrix::PFM")) { if ($args{'-matrixtype'} eq ('PWM')) { $matrix = $matrix->to_PWM(); } if ($args{'-matrixtype'} eq ('ICM')) { $matrix = $matrix->to_PWM(); } } $matrixset->add_Matrix($matrix); } return $matrixset; } sub store_MatrixSet { #DONE a wrapper around store_Matrix (which also can take an array of matrices, so utility only my ($self, $matrixset) = @_; my $it = $matrixset->Iterator(); while (my $matrix_object = $it->next) { # do whatever you want with individual matrix objects $self->store_Matrix( $matrix_object); } } =head2 delete_Matrix_having_ID Title : delete_Matrix_having_ID Usage : $db->delete_Matrix_with_ID('M00045.1'); Function: Deletes the matrix having the given ID from the database Returns : 0 on success; $@ contents on failure (this is too C-ike and may change in future versions) Args : (ID) A string. Has to be a matrix ID with version suffix in JASPAR6. Comment : Yeah, yeah, 'delete_Matrix_having_ID' is a stupid name for a method, but at least it should be obviuos what it does. =cut sub delete_Matrix_having_ID { my ($self, @IDs) = @_; # this has to be versioned IDs foreach my $ID (@IDs) { my ($base_id, $version) = split(/\./, $ID); unless ($version) { warn "You have supplied a non-versioned matrix ID to delete. Skipping $ID "; return 0; } # get relevant internal ID my ($int_id) = $self->_get_internal_id($base_id, $version); eval { my $q_ID = $self->dbh->quote($int_id); foreach my $table ( qw (MATRIX_DATA MATRIX MATRIX_SPECIES MATRIX_PROTEIN MATRIX_ANNOTATION) ) { $self->dbh->do("DELETE from $table where ID=$q_ID"); } }; } return $@; } =head2 get_TFFM_by_ID Title : get_TFFM_by_ID Usage : my $tffm = $db->get_TFFM_by_ID('TFFM0001'); Function: fetches TFFM data under the given ID from the database and returns a TFBS::TFFM object. Returns : a TFBS::TFFM object Args : TFFM ID TFFM_ID id is a string which refers to the stable JASPAR TFFM ID (usually something like "TFFM0001") with or without version numbers. "TFFM0001" will give the latest version on TFFM0001, while "TFFM0001.2" will give the second version, if existing. Warnings will be given for non-existing TFFMs. =cut sub get_TFFM_by_ID { my ($self, $id) = @_; # id is a stable ID with possible version number unless (defined $id) { $self->throw("No ID passed to get_TFFM_by_ID"); } # separate stable ID and version number my ($base_ID, $version) = split(/\./, $id); # latest version per default $version = $self->_get_TFFM_latest_version($base_ID) unless $version; # get internal ID - also a check for validity my $int_id = $self->_get_TFFM_internal_id($base_ID, $version); # get TFFM using internal ID my $tffm = $self->_get_TFFM_by_int_id($int_id); return ($tffm); } =head2 get_TFFM_by_matrix_ID Title : get_TFFM_by_matrix_ID Usage : my $tffm = $db->get_TFFM_by_matrix_ID('MA0001.1'); Function: fetches TFFM data under related to the given matrix ID from the database and returns a TFBS::TFFM object. Returns : a TFBS::TFFM object Args : Matrix ID Matrix ID id is a string which refers to the stable JASPAR matrix ID (usually something like "MA0148.3"). Note that this *should* be a fully qualified matrix ID (with version) as the TFFM is related to a specific version of a matrix. If no matrix version is given, the latest version of the matrix is retrieved and the corresponding TFFM for that matrix version is retrieved (could be no TFFM). In general only a few matrices have associated TFFMs so in many cases no TFFM will be retrieved. In these cases we return undef. =cut sub get_TFFM_by_matrix_ID { # id is a fully qualified matrix stable ID including version number my ($self, $matrix_id) = @_; unless (defined $matrix_id) { $self->throw("No matrix ID passed to get_TFFM_by_matrix_ID"); } # separate matrix stable ID and version number my ($matrix_base_id, $matrix_version) = split(/\./, $matrix_id); # latest matrix version per default $matrix_version = $self->_get_latest_version($matrix_base_id) unless $matrix_version; my $sth = $self->dbh->prepare( qq! SELECT BASE_ID, VERSION, NAME, LOG_P_1ST_ORDER, LOG_P_DETAILED, EXPERIMENT_NAME FROM TFFM WHERE MATRIX_BASE_ID = "$matrix_base_id" AND MATRIX_VERSION = "$matrix_version" ! ); $sth->execute(); my ($base_id, $version, $name, $log_p_1st_order, $log_p_detailed, $exp_name) = $sth->fetchrow_array(); my $tffm; if ($base_id) { eval { $tffm = TFBS::TFFM->new( -ID => "$base_id.$version", -name => $name, -log_p_1st_order => $log_p_1st_order, -log_p_detailed => $log_p_detailed, -experiment_name => $exp_name, # # OR we could retrieve the matrix and set the corresponding # attribute # -matrix_ID => "$matrix_base_id.$matrix_version" ) }; if ($@) { $self->throw($@); } # # Instead of storing the matrix ID, get the related matrix and store in # the matrix attribute. # # my $matrix = $self->get_Matrix_by_ID( # "$matrix_base_id.$matrix_version", 'PFM' # ); # # $tffm->matrix($matrix); # } else { warn "No TFFM exists for matrix '$matrix_base_id.$matrix_version'"; } return $tffm; } =head2 get_TFFM_by_name Title : get_TFFM_by_name Usage : my $tffm = $db->get_TFFM_by_name('HNF-1'); Function: fetches TFFM data under the given name from the database and returns a TFBS::TFFM object Returns : a TFBS::TFFM object Args : A TFFM name - the name of the transcription factor for which this TFFM was modelled. This is the same as the name of the matrix used to train the TFFM. Warning : According to the current JASPAR data model, name is not necessarily a unique identifier. Also, names change over time. In the case where there are several TFFMs with the same name in the database, the function fetches the first one and prints a warning on STDERR. You've been warned. Some matrices have multiple versions. The function will return the latest version. For specific versions, use get_TFFM_by_ID($ID.$version) =cut sub get_TFFM_by_name { my ($self, $name) = @_; unless (defined $name) { $self->throw("No name passed to get_TFFM_by_name."); } # sanity check: are there many different stable IDs with same name? my $sth = $self->dbh->prepare( qq!SELECT distinct BASE_ID FROM TFFM WHERE NAME="$name"! ); $sth->execute(); my (@stable_ids) = $sth->fetchrow_array(); my $L = scalar @stable_ids; $self->warn("There are $L distinct stable IDs with name '$name'") if scalar $L > 1; return $self->get_TFFM_by_ID($stable_ids[0]); } ######################################################################### # PRIVATE METHODS ######################################################################### sub _new { #PROBABLY OK my ($caller, %args) = @_; my $class = ref $caller || $caller; my $self = bless {}, $class; my ($connectstring, $user, $password); if ($args{'-connect'} and (ref($args{'-connect'}) eq "ARRAY")) { ($connectstring, $user, $password) = @{$args{'-connect'}}; } elsif ($args{'-create'} and (ref($args{'-create'}) eq "ARRAY")) { return $caller->create(@{-args {'create'}}); } else { ($connectstring, $user, $password) = (DEFAULT_CONNECTSTRING, DEFAULT_USER, DEFAULT_PASSWORD); } $self->dbh(DBI->connect($connectstring, $user, $password, {mysql_enable_utf8 => 1})); return $self; } sub _store_matrix_data { # DONE my ($self, $pfm, $int_id, $ACTION) = @_; my @base = qw(A C G T); my $matrix = $pfm->matrix(); my $type; my $sth = $self->dbh->prepare(q! INSERT INTO MATRIX_DATA VALUES(?,?,?,?) !); for my $i (0 .. 3) { for my $j (0 .. ($pfm->length - 1)) { $sth->execute( $int_id, $base[$i], $j + 1, $matrix->[$i][$j] ) or $self->throw("Error executing query."); } } } sub _store_matrix { #DONE # creation of the matrix will also give an internal unique ID # (incremental int) which will be returned to use for the other tables my ($self, $pfm, $ACTION) = @_; # # Added check that the version is not already stored as part of the ID. # Also added a more informative insertion exception message. # DJA 2016/08/26 # my $id = $pfm->ID; my $base_id; my $version; if ($id =~ /^(\S+)\.(\d+)/) { $base_id = $1; $version = $2; } else { $base_id = $id; } unless ($version) { # Get collection and version from the matrix tags $version = $pfm->{'tags'}{'version'}; } # will warn but not die if version is missing: will assume 1 unless ($version) { warn "WARNING: Lacking version number for " . $pfm->ID . ". Setting version=1"; $version = 1; } my $collection = $pfm->{'tags'}{'collection'}; unless ($collection) { warn "WARNING: Lacking collection name for " . $pfm->ID . ". Setting collection to an empty string. You probably do not want this"; $collection = ''; } # sanity check: do we already have this cobination of base ID and version? # If we do, die my $sth = $self->dbh->prepare( qq! select count(*) from MATRIX where VERSION=$version and BASE_ID= "$base_id" and collection="$collection" ! ); $sth->execute; my ($sanity_count) = $sth->fetchrow_array; if ($sanity_count > 0) { warn "WARNING: Database input inconsistency: You have already have $sanity_count $base_id matrices of version $version in collection $collection. Terminating program"; die; } # insert data $sth = $self->dbh->prepare( q! INSERT INTO MATRIX VALUES(?,?,?,?,?) ! ); # update next sth with actual version and collection: DO $sth->execute(0, $collection, $pfm->ID, $version, $pfm->name) or $self->throw( sprintf("Error inserting matrix %s as %s.%d to %s collection", $pfm->name, $pfm->ID, $version, $collection) ); # get the actual (new) iternal ID my $int_id = $self->dbh->{q{mysql_insertid}}; return $int_id; } sub _store_matrix_annotation { # DONE # # this is for tag-value items that are not one-to-many (so, not species # and not acc) # # We do need to store multiple values of class and family. In the case # of profiles representing dimers, the TFs making up the dimer may have # different TF classes and families. It also seemed abitrary to have a # primary key constraint on the ID + TAG. This has been removed. # # Added more informative messages to insertion failure exceptions. # Modified split to also handle spaces around comma. # DJA 2015/08/26 # my ($self, $pfm, $int_id, $ACTION) = @_; my $sth = $self->dbh->prepare( q! INSERT INTO MATRIX_ANNOTATION (ID, tag, val) VALUES(?,?,?) ! ); # get all tags # but skip out collection or version as we already have those in the # MATRIX table # special handling for class which mighht have a true slot my %tags = $pfm->all_tags(); if (defined($pfm->{class})) { $tags{class} = $pfm->{class}; } foreach my $tag (keys %tags) { next if $tag eq "collection"; next if $tag eq "version"; next if $tag eq "species"; # # The 'acc' tag was commented out in JASPAR6. Not sure the reasoning # as this stores the protein accession (UniProt IDs) which are # stored in the MATRIX_PROTEIN table and therefore should NOT be # stored in the MATRIX_ANNOTATION table and therefore *should* be # skipped here. # DJA 2015/08/26 # next if $tag eq "acc"; # next if $tag eq "class"; #$sth->execute($int_id, $tag, ($tags{$tag} or ""),) # or $self->throw("Error executing query"); # # Since we may have multiple values stored in some tags we need to # handle this. Assume that they may be stored either as array # references or as strings with comma separators. # DJA 2015/08/26 # my $vals = $tags{$tag}; if (ref $vals eq 'ARRAY') { foreach my $val (@$vals) { $val =~ s/^\s+//; $val =~ s/\s+$//; $sth->execute($int_id, $tag, $val) or $self->throw( "Error inserting tag/value pair ('$tag', '$val')" . " into MATRIX_ANNOTATION" ); } } else { foreach my $val (split(/\s*\,\s*/, $vals)) { $val =~ s/^\s+//; $val =~ s/\s+$//; $sth->execute($int_id, $tag, $val) or $self->throw( "Error inserting tag/value pair ('$tag', '$val')" . " into MATRIX_ANNOTATION" ); } } } } sub _store_matrix_species { # DONE #these are for species IDs - can be several # these are taken from the tag "species" # if that tag is a reference to an array we walk over the array # if it is a comma-separated string we split the string # # Added exception handling/messages for insertion failure. # Modified split to also handle spaces around comma. # DJA 2015/08/26 # my ($self, $pfm, $int_id, $ACTION) = @_; my $sth = $self->dbh->prepare( q! INSERT INTO MATRIX_SPECIES VALUES(?,?) ! ); #sanity check: are there any species? Its ok not to have it. return () unless $pfm->{'tags'}{'species'}; #is the species a string or an arrayref? if (ref($pfm->{'tags'}{'species'}) eq 'ARRAY') { # walkthru array foreach my $species (@{$pfm->{'tags'}{'species'}}) { $sth->execute($int_id, $species) or $self->throw( "Error inserting species '$species' for matrix $int_id)" ); } } else { # split and walk thru foreach my $species (split(/\s*\,\s*/, $pfm->{'tags'}{'species'})) { $species =~ s/^\s+//; $sth->execute($int_id, $species) or $self->throw( "Error inserting species '$species' for matrix $int_id)" ); } } } sub _store_matrix_acc { # DONE #these are for protein accession numbers - can be several # these are taken from the tag "acc" # if that tag is a reference to an array we walk over the array # if it is a comma-separated string we split the string # # For some reason the MATRIX_PROTEIN table had a primary key constraint # on the matrix ID field, preventing multiple proteins from being stored # here. That was presumably an oversight or otherwise an out of date table # definition. # # Also added more informative insertion exception messages. # DJA 2016/08/26 # my ($self, $pfm, $int_id, $ACTION) = @_; my $sth = $self->dbh->prepare( q! INSERT INTO MATRIX_PROTEIN VALUES(?,?) ! ); #sanity check: are there any accession numbers? Its ok not to have it. return () unless $pfm->{'tags'}{'acc'}; #is the protein ID a string or an arrayref? if (ref($pfm->{'tags'}{'acc'}) eq 'ARRAY') { # walkthru array foreach my $acc (@{$pfm->{'tags'}{'acc'}}) { $acc =~ s/\s//g; $sth->execute($int_id, $acc) or $self->throw( "Error inserting protein '$acc' for matrix $int_id)" ); } } else { # split and walk thru foreach my $acc (split(/\,/, $pfm->{'tags'}{'acc'})) { $acc =~ s/\s//g; $sth->execute($int_id, $acc) or $self->throw( "Error inserting protein '$acc' for matrix $int_id)" ); } } } #when creating: try to support arbitrary tags sub _create_tables { # DONE # utility function # If you want to change the databse schema, # this is the right place to do it # # Changed the primary key constraint on MATRIX_ANNOTATION to a simple # key. There are tags for which we do want to have multiple entries, e.g. # in the case of a dimer profiles the two TFs making up the dimer may # have a different TF class and/or family therefore we need to store more # than one record with tag 'class' or tag 'family' for the same matrix # ID. # # Also added key on ID to MATRIX_PROTEIN and MATRIX_SPECIES which seemed # to be missing. # # We may also want to set the charset/collation to utf8/utf8_unicode_ci # either at the DB level or to the individual table definitions # (particularly for the MATRIX_ANNOTATION table) to handle non-Latin # characters correctly. # # DJA 2015/08/26 # my $dbh = shift; my @queries = ( q! CREATE TABLE MATRIX ( ID INT(11) NOT NULL AUTO_INCREMENT, COLLECTION VARCHAR (16) DEFAULT '', BASE_ID VARCHAR (16) DEFAULT '' NOT NULL , VERSION TINYINT(4) DEFAULT 1 NOT NULL , NAME VARCHAR (255) DEFAULT '' NOT NULL, PRIMARY KEY (ID)) !, q! CREATE TABLE MATRIX_DATA ( ID INT(11) NOT NULL, row VARCHAR(1) NOT NULL, col TINYINT(3) UNSIGNED NOT NULL, val float(10,3), PRIMARY KEY (ID, row, col)) !, q! CREATE TABLE MATRIX_ANNOTATION ( ID INT(11) NOT NULL, TAG VARCHAR(255) DEFAULT '' NOT NULL, VAL varchar(255) DEFAULT '', KEY (ID, TAG)) !, q! CREATE TABLE MATRIX_SPECIES ( ID INT(11) NOT NULL, TAX_ID VARCHAR(255) DEFAULT '' NOT NULL KEY (ID)) !, q! CREATE TABLE MATRIX_PROTEIN ( ID INT(11) NOT NULL, ACC VARCHAR(255) DEFAULT '' NOT NULL KEY (ID)) !, q! CREATE TABLE TFFM ( ID int(11) NOT NULL auto_increment, BASE_ID varchar(16) NOT NULL, VERSION tinyint(4) NOT NULL, MATRIX_BASE_ID varchar(16) NOT NULL, MATRIX_VERSION tinyint(4) NOT NULL, NAME varchar(255) NOT NULL, LOG_P_1ST_ORDER float default NULL, LOG_P_DETAILED float default NULL, EXPERIMENT_NAME varchar(255) default NULL, PRIMARY KEY (ID), KEY BASE_ID (BASE_ID, VERSION), KEY MATRIX_BASE_ID (MATRIX_BASE_ID, MATRIX_VERSION) ! ); foreach my $query (@queries) { $dbh->do($query) or die("Error executing the query: $query\n"); } } sub _get_matrixstring { #DONE my ($self, $ID) = @_; #my %dbname = (PWM => 'pwm', PFM => 'raw', ICM => 'info'); #unless (defined $dbname{$mt}) { #$self->throw("Unsupported matrix type: ".$mt); #} my $sth; my $qID = $self->dbh->quote($ID); my $matrixstring = ""; foreach my $base (qw(A C G T)) { $sth = $self->dbh->prepare( "SELECT val FROM MATRIX_DATA WHERE ID=$qID AND row='$base' ORDER BY col" ); $sth->execute; $matrixstring .= join(" ", (map {$_->[0]} @{$sth->fetchall_arrayref()})) . "\n"; } $sth->finish; return undef if $matrixstring eq "\n" x 4; return $matrixstring; } sub _get_latest_version { #DONE my ($self, $base_ID) = @_; # SELECT VERSION FROM MATRIX WHERE BASE_ID=? ORDER BY VERSION DESC LIMIT 1 my $sth = $self->dbh->prepare( qq!SELECT VERSION FROM MATRIX WHERE BASE_ID="$base_ID" ORDER BY VERSION DESC LIMIT 1! ); $sth->execute; my ($latest) = $sth->fetchrow_array(); return ($latest); } sub _get_internal_id { #DONE # picks out the internal id for a a stable id+ version. Also checks if this cobo exists or not my ($self, $base_ID, $version) = @_; # SELECT ID FROM MATRIX WHERE BASE_ID=? and VERSION=? my $sth = $self->dbh->prepare( qq!SELECT ID FROM MATRIX WHERE BASE_ID="$base_ID" AND VERSION="$version"! ); $sth->execute; my ($int_id) = $sth->fetchrow_array(); return ($int_id); } sub _get_Matrix_by_int_id { #done my ($self, $int_id, $mt) = @_; my $matrixobj; $mt = 'PFM' unless $mt; # get the matrix as a string my $matrixstring = $self->_get_matrixstring($int_id) || return undef; #get remaining data in the matrix table: name, collection my $sth = $self->dbh->prepare( qq!SELECT BASE_ID,VERSION, COLLECTION,NAME FROM MATRIX WHERE ID="$int_id"! ); $sth->execute(); my ($base_ID, $version, $collection, $name) = $sth->fetchrow_array(); # jsp6 # get species ##$sth=$self->dbh->prepare(qq!SELECT TAX_ID FROM MATRIX_SPECIES WHERE ID="$int_id"!); $sth = $self->dbh->prepare( qq!SELECT GROUP_CONCAT(TAX_ID SEPARATOR ', ') as TAX_ID FROM MATRIX_SPECIES WHERE ID="$int_id"! ); $sth->execute(); my @tax_ids; while (my ($res) = $sth->fetchrow_array()) { my @res_v = split(/,/, $res); my @res_v2 = grep(s/^\s*(.*)\s*$/\1/g, @res_v); push(@tax_ids, @res_v2); } # jsp6 # get acc ##$sth=$self->dbh->prepare(qq!SELECT ACC FROM MATRIX_PROTEIN WHERE ID="$int_id"!); $sth = $self->dbh->prepare( qq!SELECT GROUP_CONCAT(ACC SEPARATOR ', ') as ACC FROM MATRIX_PROTEIN WHERE ID="$int_id"! ); $sth->execute(); my @accs; while (my ($res) = $sth->fetchrow_array()) { my @res_v = split(/,/, $res); my @res_v2 = grep(s/^\s*(.*)\s*$/\1/g, @res_v); push(@accs, @res_v2); } # jsp6 # get remaining annotation as tags, form ANNOTATION table my %tags; $sth = $self->dbh->prepare( qq{SELECT TAG, VAL FROM MATRIX_ANNOTATION WHERE ID = "$int_id" }); $sth->execute(); ## my @key_to_split=("acc", "medline", "pazar_tf_id"); #if acc in MATRIX_ANNOTATION #my @key_to_split=("medline", "pazar_tf_id"); # # Added 'class' and 'family' to keys to split. Since we have dimers that # may have different classes / families. Previously stored in DB as comma # separated lists, now store as separate records. # XXX But this breaks the interface code! FIXME XXX # DJA 2015/09/10 # #my @key_to_split = ("medline", "pazar_tf_id", "tfbs_shape_id", "tfe_id", "class", "family"); my @key_to_split = ("class", "family", "medline", "pazar_tf_id", "tfbs_shape_id", "tfe_id"); # # See FIXME comment below. # DJA 2015/09/14 # #foreach my $key (@key_to_split) { # $tags{$key} = ['-']; #} # # Fixed so that we can have multiple comma separated values, values in # separate rows of the table or a combination thereof. We really should # try to get away from having to specify which keys can be split (should # be handled more gracefully). # DJA 2015/09/14 # my $vals; while (my ($tag, $val) = $sth->fetchrow_array()) { $vals = []; if ($tag ~~ @key_to_split) { my @val_v = split(/,/, $val); my @val_v2 = grep(s/^\s*(.*)\s*$/\1/g, @val_v); #push(@$vals, @val_v2); #$tags{$tag} = $vals; push @{$tags{$tag}}, @val_v2; } else { $tags{$tag} = $val; } } # # XXX FIXME # This really doesn't belong here. It is done for the purposes of the web # interface but this is DB code which should not presume how the returned # data is going to be used. That should be handled in the JASPAR web code # modules. JASPAR web code currently expects all keys to split to exist! # DJA 2015/09/14 # XXX FIXME # foreach my $key (@key_to_split) { $tags{$key} = ['-'] unless $tags{$key}; } # jsp6 $tags{'collection'} = $collection; $tags{'species'} = \@tax_ids; # as array reference instead of strigifying $tags{'acc'} = \@accs; # same, if acc MATRIX_PROTEIN # my $class = $tags{'class'}; delete($tags{'class'}); # eval( "\$matrixobj = TFBS::Matrix::PFM->new" . ' ( -ID => "$base_ID.$version", -name => $name, -class => $class, -tags => \%tags, -matrixstring => $matrixstring # FIXME - temporary );' ); if ($@) { $self->throw($@); } # warn $int_id, "\t", ref($matrixobj); return ($matrixobj->to_PWM) if $mt eq "PWM"; return ($matrixobj->to_ICM) if $mt eq "ICM"; return ($matrixobj); # default PFM } ##jsp6 sub _get_IDlist_by_query { #needs cleanup. NOT for the faint-hearted. my ($self, %args) = @_; warn '_get_IDlist_by_query | $self || ', $self; warn '_get_IDlist_by_query | %args || ', %args; # called by get_MatrixSet # warn $args{"-collection"}; $args{'-collection'} = 'CORE' unless $args{'-collection'}; # returns a set of internal IDs with whicj to get the actual matrices # current idea: # 1: first catch non-tag things like collection, name and version, species # makw one query for these if they are named and check the IDs for "latest" unless requested not to. # these are AND statements # 2:then do the rest on tag level: # to be able to do this with actual and tattemnet innthe tag table, we do an inner join query, which is kept separate just for convenice # we then intersect 1 and 2 # 3: then do matrix-based features such as ic, with, number of sites etc, for the surviving matrices. This shold happen in the get_matrixset part my @int_ids_to_return; ## jsp6 - autosearch if ($args{'-auto'}) { ##my $sth=$self->dbh->prepare (qq!SELECT ID FROM MATRIX WHERE BASE_ID=?!); my $sth = $self->dbh->prepare( qq!SELECT U.ID FROM (SELECT ID, BASE_ID as VAL FROM MATRIX UNION ALL SELECT ID, NAME as VAL FROM MATRIX UNION ALL SELECT ID, ACC as VAL FROM MATRIX_PROTEIN UNION ALL SELECT ID, TAX_ID as VAL FROM MATRIX_SPECIES UNION ALL SELECT ID, SPECIES as VAL FROM MATRIX_SPECIES,TAX WHERE MATRIX_SPECIES.TAX_ID=TAX.TAX_ID UNION ALL SELECT ID, NAME as VAL FROM MATRIX_SPECIES,TAX_EXT WHERE MATRIX_SPECIES.TAX_ID=TAX_EXT.TAX_ID AND MATRIX_SPECIES.TAX_ID=9606 UNION ALL SELECT ID, VAL as VAL FROM MATRIX_ANNOTATION) AS U WHERE LOWER(`VAL`) LIKE LOWER(?)! ); warn '_get_IDlist_by_query | $sth || ', $sth; foreach my $stID (@{$args{'-auto'}}) { warn '_get_IDlist_by_query | $stID || ', $stID; my ($stable_ID, $version) = split(/\./, $stID) ; # ignore vesion here, this is a stupidity filter #$sth->execute($stable_ID); $sth->execute("%" . $stable_ID . "%"); while (my ($int_id) = $sth->fetchrow_array()) { warn '_get_IDlist_by_query | $int_id || ', $int_id; push(@int_ids_to_return, $int_id); } } return \@int_ids_to_return; } # should redo so that matrix_annotation queries are separate, with an intersect in the end #special case 1: get ALL matrices. Higher priority than all if ($args{'-all'}) { my $sth = $self->dbh->prepare(qq!SELECT ID FROM MATRIX!); $sth->execute(); my @a; while (my ($i) = $sth->fetchrow_array()) { push(@a, $i); } return \@a; } # ids: special case2 which is has higher priority than any other except the above (ignore all others if ($args{'-ID'}) { # these might be either stable IDs or stableid.version. # if just stable ID and if all_versions==1, take all versions, otherwise the latest if ($args{-all_versions}) { my $sth = $self->dbh->prepare( qq!SELECT ID FROM MATRIX WHERE BASE_ID=?!); foreach my $stID (@{$args{'-ID'}}) { my ($stable_ID, $version) = split(/\./, $stID) ; # ignore vesion here, this is a stupidity filter $sth->execute($stable_ID); while (my ($int_id) = $sth->fetchrow_array()) { push(@int_ids_to_return, $int_id); } } } else { # only the lastest version, or the requested version foreach my $stID (@{$args{'-ID'}}) { #warn $stID; my ($stable_ID, $version) = split(/\./, $stID); $version = $self->_get_latest_version($stable_ID) unless $version; my $int_id = $self->_get_internal_id($stable_ID, $version); push(@int_ids_to_return, $int_id) if $int_id; } } return \@int_ids_to_return; } my @tables = ("MATRIX M"); my @and; # in matrix table: collection, if ($args{-collection}) { my $q = ' (COLLECTION='; if (ref $args{-collection} eq "ARRAY") { # so, possibly several my @a; foreach (@{$args{-collection}}) { push(@a, "\"$_\""); } $q .= join(" or COLLECTION=", @a); } else { # just one - typical usage $q .= "\"$args{-collection}\""; } $q .= " ) "; push(@and, $q); } # in matrix table: names. Is something that is basically only used from the web interface # typically used by the get_matrix_by_name function instead if ($args{-name}) { my $q = ' (NAME='; if (ref $args{-name} eq "ARRAY") { # so, possibly several my @a; foreach (@{$args{-name}}) { push(@a, "\"$_\""); } $q .= join(" or NAME=", @a); } else { # just one - typical usage $q .= "\"$args{-name}\""; } $q .= " ) "; push(@and, $q); } # in species table: tax.id: possibly many species with OR in between if ($args{-species}) { push(@tables, "MATRIX_SPECIES S"); my $q = " M.ID=S.ID and (TAX_ID= "; if (ref $args{-species} eq "ARRAY") { # so, possibly several my @a; foreach (@{$args{-species}}) { push(@a, "\"$_\""); } $q .= join(" or TAX_ID=", @a); } else { # just one - typical usage $q .= "=\"$args{-species}\""; } $q .= ") "; push(@and, $q); } # TAG_BASED # an internal join query:should be able to handle up to 26 tags-value combos with ANDS in between # Very ugly code ahead: my (@inner_tables, @internal_ands1, @internal_ands2); my $int_counter = 0; # for keeping track of names; my @alpha = ("a" .. "z"); my %arrayref; foreach my $key (keys %args) { next if $key eq "-min_ic"; next if $key eq "-matrixtype"; next if $key eq "-species"; next if $key eq "-collection"; next if $key eq "-all_versions"; next if $key eq "-all"; next if $key eq "-ID"; next if $key eq "-length"; next if $key eq "-name"; my $oldkey = $key; $key =~ s/-//; $arrayref{$key} = $args{$oldkey}; } if (%arrayref) { # get an internal name for the table push(@internal_ands2, " M.ID=a.ID "); my @a; foreach my $key (keys %arrayref) { my $tname = $alpha[$int_counter]; push(@inner_tables, "MATRIX_ANNOTATION $tname"); push(@internal_ands1, $alpha[$int_counter] . ".ID=" . $alpha[$int_counter - 1] . ".ID") unless $int_counter == 0; $int_counter++; # is the thing aupplied an array reference in inteslf: make an "or" query from that if (ref $arrayref{$key} eq "ARRAY") { my @b; foreach (@{$arrayref{$key}}) { push(@b, $self->dbh->quote($_)); } my $orstring = join(" or $tname.VAL=", @b); push(@a, "($tname.TAG=\"$key\" AND ($tname.VAL=$orstring))"); } #or not else { push(@a, "($tname.TAG=\"$key\" AND $tname.VAL=\"$arrayref{$key}\")" ); } } my $s = " ( " . join(" AND ", @a) . ")"; push(@internal_ands2, $s); } my $qq = "SELECT distinct(M.ID) from " . join(",", (@tables, @inner_tables)) . " where" . join(" AND ", (@and, @internal_ands1, @internal_ands2)); # warn $qq; #do actual mammoth query,and check for latest matrix my $sth = $self->dbh->prepare($qq); $sth->execute(); my @r; while (my ($int_id) = $sth->fetchrow_array) { if ($args{-all_versions}) { push(@r, $int_id); } else { # is latest? push(@r, $int_id) if ($self->_is_latest_version($int_id) == 1); } } warn "Warning: Zero matrices returned with current critera" unless scalar @r; return \@r; } ## jsp6 - checkpoint sub _is_latest_version { # is a particular internal ID representingthe latest matrix (collapse on base ids) my ($self, $int_id) = @_; my $sth = $self->dbh->prepare( qq! select count(*) from MATRIX where BASE_ID= (SELECT BASE_ID from MATRIX where ID=$int_id) AND VERSION>(SELECT VERSION from MATRIX where ID=$int_id) ! ); $sth->execute(); my ($count) = $sth->fetchrow_array(); return (1) if $count == 0; # no matrices with higher version ID and same base id return (0); } sub _get_TFFM_latest_version { my ($self, $base_ID) = @_; # SELECT VERSION FROM TFFM WHERE BASE_ID=? ORDER BY VERSION DESC LIMIT 1 my $sth = $self->dbh->prepare( qq!SELECT VERSION FROM TFFM WHERE BASE_ID="$base_ID" ORDER BY VERSION DESC LIMIT 1! ); $sth->execute; my ($latest) = $sth->fetchrow_array(); return ($latest); } sub _get_TFFM_internal_id { # picks out the internal id for a stable id version. Also checks if this # combo exists or not my ($self, $base_ID, $version) = @_; # SELECT ID FROM TFFM WHERE BASE_ID=? and VERSION=? my $sth = $self->dbh->prepare( qq!SELECT ID FROM TFFM WHERE BASE_ID="$base_ID" AND VERSION="$version"! ); $sth->execute; my ($int_id) = $sth->fetchrow_array(); return ($int_id); } sub _get_TFFM_by_int_id { #done my ($self, $int_id) = @_; my $tffm; my $sth = $self->dbh->prepare( qq! SELECT BASE_ID, VERSION, MATRIX_BASE_ID, MATRIX_VERSION, NAME, LOG_P_1ST_ORDER, LOG_P_DETAILED, EXPERIMENT_NAME FROM TFFM WHERE ID = "$int_id" ! ); $sth->execute(); my ($base_id, $version, $matrix_base_id, $matrix_version, $name, $log_p_1st_order, $log_p_detailed, $exp_name) = $sth->fetchrow_array(); my $tffm; eval { $tffm = TFBS::TFFM->new( -ID => "$base_id.$version", -name => $name, -log_p_1st_order => $log_p_1st_order, -log_p_detailed => $log_p_detailed, -experiment_name => $exp_name, # # OR we could retrieve the matrix and set the corresponding # attribute # -matrix_ID => "$matrix_base_id.$matrix_version" ) }; if ($@) { $self->throw($@); } # # Instead of storing the matrix ID, get the related matrix and store in # the matrix attribute. # # my $matrix = $self->get_Matrix_by_ID( # "$matrix_base_id.$matrix_version", 'PFM' # ); # # $tffm->matrix($matrix); # return $tffm; # default PFM } sub _is_TFFM_latest_version { my ($self, $int_id) = @_; my $sth = $self->dbh->prepare( qq! select count(*) from TFFM where BASE_ID = (SELECT BASE_ID from TFFM where ID = $int_id) AND VERSION > (SELECT VERSION from TFFM where ID = $int_id) ! ); $sth->execute(); my ($count) = $sth->fetchrow_array(); # no TFFMs with higher version ID and same base id return (1) if $count == 0; return (0); } sub DESTROY { #OK $_[0]->dbh->disconnect() if $_[0]->dbh; }