% 10pt is the smallest font for article
\documentclass{article}

\usepackage{graphicx}
\usepackage{epsf}
% \usepackage{a4}
\usepackage{palatino}
\usepackage{euler}
\newcommand {\atilde} {$_{\char '176}$} % tilde(~) character

\title{Tutorial: Validation with Coot}

% \author{CCP4 Workshop New Delhi 2010}
% \author{BCA/CCP4 Summer School Oxford 2010}
\author{CSHL 2017}

\begin{document}
\maketitle
%\tableofcontents
%\listoffigures

\section{Preamble}

  We will validate the structure in which we fitted the
  3-aminobenzamide ligands.

\section{Difference Map Peaks}

  One of the best tools for finding problems is the difference maps
  peaks tool. So let's use that that.  

  You should be able to see the ligand sites and other interesting
  features. What the problem here? (For me it is peak 5, but if you
  have run refmac with the model from the ligand fitting, then it may
  be peak number 1.)  Fix the problem.

\section{Rotamer Analysis}
 
Another useful tool is the Rotamer Analysis.  Use that tool (in the
\textsf{Validate} menu) to find problematic residues (note: not all
big bars are necessarily wrong).

%   [ Note to demonstrator:]

%   LEU A 358 is back to front 
%   VAL A 567 is back to front

%   LYS A 263 model as an ALA
%   ILE A 289 round the wrong way

%   TRP A 427 round the wrong way

\section{Ramachandran Plot and Difference Plots}

The Ramachandran Plot is a classic model-based validation tool.  The
Kleywegt Plot (\emph{i.e.} Ramachandran NCS difference Plot) is often
used to examine NCS-related mainchain differences.

Use these tools on this model to find Rotamers with problematic
phi/psi values.  What is the Kleywegt Plot telling you? Is there a
problem? If there is a problem, how might you fix it? (Hint is there
anything in Refinement/Regularization Control (R/RC) that might be of
use?)

\section{NCS and NCS Ghosts}

   Another way of showing NCS differences is using the NCS Ghosts, NCS
   Maps and NCS skipping.  

\subsection{NCS Skipping}

Press \texttt{"P"} to go to the nearest and then \texttt{"O"} to jump
between NCS-related models preserving the relative (NCS-compensating)
view.

\subsection{NCS Ghosts}

\textsf{Draw $\rightarrow$ NCS Ghosts Control\ldots} Then for the
chosen molecule turn on the NCS ghosts.

\subsection{NCS Maps}

NCS maps (\textsf{Calculate $\rightarrow$ NCS maps}) are useful
addition to the NCS Ghosts because they show the maps corresponding
the the NCS ghost molecules in the context of the NCS master
(typically "A") chain.

 
\section{Sequence Validation}

Sometime people make sequence-based errors in building models.  Coot
has a tool that allows you to compare the sequence in the model with a
reference sequence.

\textsf{Validate $\rightarrow$ Alignment vs. PIR}

Read in the sequence file and press ``OK''.

Is there a sequence problem?  How might you fix it?

\section{Freestyle...}

Use the EDS to download the structure and data/map for \texttt{1BAV}.
Use the above tools to find any issues and correct them.

How about \texttt{1H4P}?  Any issues there?  (You can change the
refinement configurations by pressing the R/RC button (Hint: Read the tooltips).)

%% 1g72 A269 has a real non-pro cis-peptide
%% 5upv A312 has a real non-pro cis-peptide

% Dale Trp 1qw9

% 3aj8 - outliers in cablam - also CBeta deviations for PROlines

And how about \texttt{2XDE}? % example of checking by NCS jumping (top
                             % peak), pepflip and LEU 6B.
(Hint: Diff map peaks, NCS jumping).

And how about \texttt{1QW9}?

Check for more weird waters (overlaps) - water on water overlaps - might be
something else?

%% Glycerol

And \texttt{1QEX}?  Problems?

And how about \texttt{3L0F}? (Easy to spot, non-trivial to
fix.) % example of ligand with incorrect restraints

Another: \texttt{3F1L}

% A832 -12 difference map peak

Another: \texttt{1BJI}

% 3eni (new) vs 1m50 (old)

% 3f1l high res, some model building problems, wrong sidechains, including a leu.

Another: \texttt{3F1L}

Another: \texttt{1UG6} (noted by Dale Tronrud)

%% coot-download/4mt6.ent use this for competition?  Can you beat the deposited structure?
%% Need to find the correct sequence though.
%%
Another: \texttt{4Q86} (look at the difference map)

% misfit residues and ligand difference map (too strict dictionary)
Another: \texttt{2XIR}

% 3b3q waters without hydrogen bonds: Robbie J

% Carbohydrate building - not really validation.
%   
%   o 4gos is great for carbohydrate building (Andre L. example).  Also
%     the BMA fails to build in corectly.  It's a ligand LSQ fit
%     problem, I think. When this works, put it in the GUI.

% 3fvl (from Robbie) Cis-peptides, ligands.

Others (via Robbie Joosten): \texttt{1CYG}, \texttt{3FVL}, \texttt{3ABA}.
5ONU (described in the CaBLAM/Undowswer paper, Prisant et al (2019)).

% 467A in 1hi8. Peptide flip - in the middle of a helix!

% and ILE misfitted
% Another: 1cyg (from Robbie Joosten)

% I found this one: B141 in 1qx5 (non-pro cis)


% 1mxi is knotted: Toni


\section*{Bottom line:}

I have little patience for those pointing out that there are
structures with problems in the PDB. Yes, of course there are. All
structures have "errors" to some lesser or greater extent (mine
included).

The point of this excercise is not to pick holes in other people's
structures - it is to illustrate what problems look like and that
using modern graphics software (and in particular, Coot) such problems
can be readily identified and fixed.

% It's an opportunity to learn from and improve the software.

\begin{quotation}
  ``At least one person should look at the map\ldots'' 

  -- Dale Tronrud\footnote{I believe}
\end{quotation}

\end{document}