zormit / compbio-winter14 Goto Github PK

noisy constraint analysis
=========================

Configuration
-------------

An example config can be found in `paths.cfg`. You most certaintly have
to adapt it, such that your rosetta binaries and database are linked.
Also make sure that for a protein with <pdb-id>, the input directory contains
a directory with the same name and the following files:

* Sequence File: <pdb-id>.fasta
* Native Protein: <pdb-id>.pdb
* Fragment Files: <pdb-id>.200.9mers, <pdb-id>.200.3mers
* Noisy Constraint File: <pdb-id>_contact_constraints.txt

Example structure of my input directory:

    $ ls Data/input/**
    Data/input/2h3jA:
    2h3jA.200.3mers  2h3jA.200.9mers  2h3jA_contact_constraints.txt  2h3jA.fasta  2h3jA.pdb

    Data/input/2krkA:
    2krkA.200.3mers  2krkA.200.9mers  2krkA_contact_constraints.txt  2krkA.fasta  2krkA.pdb

The section [filename] normally does not have to be changed.


Basic Usage
-----------

To run the script, call:

    ./runExperiment.py <config-filename> <pdb-id>

When the run is successful, a lot of files will be generated:

1. <input-dir>/<pdb-id>/generated/ will contain a constraint-file with the
   measured native distances as a new column for each constraint/residue-pair.
2. <output-dir>/<pdb-id>/ will have a subdirectory for each prediction with.

   a) the constraint subset file, which will guide the prediction
   b) the standard rosetta output, with decoys (pdb&silentfile) and score.fsc
   c) the rescored scorefile (standard: scoreV2.fsc), with constraint-penalty-free scores.

3. <plot-dir>/<pdb-id>/ will contain lots of plots.

*Attention*: all these files will be overwritten by the scripts on a re-run.
           (except decoys. to overwrite them, delete default.out in a subdir.)


-----
Flags
-----

* `-debug/-d` runs rosetta in a debug mode, i.e. without relaxing
  and only predicting two decoys.
* `-verbose/-v` gives you more information about what happens on the console.
  but all this information and more can be found in the logfile anyway.

constraint subset generation methods
------------------------------------

The datastructure to choose a constraint is basically an adjacency matrix,
where a number >= 0 denotes a chosen residue pair. Same number means same subset.

To add other constraint subsets, simply add a matrix
to the three datastructures in `generate_constraint_subsets()`.

custom plots
------------

The plotting "system" is written in a way, that it is possible to compare different groups of subsets.
For example, to use another target or baseline, modify the string variables accordingly.