SRS for ligand-protein binding
09/27/04

Mehmet Serkan Apaydin
apaydin@cs.stanford.edu

Warning: Absolutely no warranty.

This is the alpha release of the SRS software for ligand-protein 
binding. It can find potential binding sites for a ligand
around a protein, or compute a roadmap for a ligand-protein complex
 around a given potential binding site. 
It writes the transition probabilities 
assigned to the edges of the roadmap to a file. The provided matlab
scripts read this file and compute the escape time from the given
potential binding site. 


The ligand-protein modelling software is based on Amit Singh's
LDOCK [3]. The ligand is modeled as flexible, whereas the protein 
is rigid. The software requires an energy grid file precomputed
using Delphi[1]. Sample grid files are provided. See [2] for more
details on the computation of this file.

The software samples conformations such that all nodes are within
15A cRMSD of the ligand conformation at the given potential 
binding site. The escape time refers to the average number of hops
on the roadmap it takes for the ligand to reach a conformation of
more than 10A cRMSD of its conformation at the binding site.

The potential binding sites are conformations of low energy, close 
to the protein surface, and distant from each other. The software
finds them by first sampling many conformations, then filtering 
them according to their distance from the protein surface and
their pairwise distances.


INSTALLATION/RUNNING INSTRUCTIONS:
1- to install, simply type tar xvfz srs_ldock.tgz
2- cd toRelease
3- make

To run, type:

./ldock_fineSampling complex_prefix roadmapFilename numNodes numLinksPerNode CoordinatesOfPotBindingSites potBindingSiteIndex.

e.g.,
./ldock_fineSampling complexes/1ldm/1ldm myRoadmap.txt 1000 7 complexes/1ldm/potBindingSiteIndices.txt 1

would run the program, create a roadmap around potential binding
site# 1  (the first entry in 
complexes/1ldm/potBindingSiteIndices.txt), 
sample 1000 nodes, connects each node to at least 7
other nodes. 7 because the ligand for 1ldm has 7 degrees of
freedom (DoF), so one link per DoF.

Sample files for the "1ldm" complex are provided.

potBindingSiteIndex is a digit between 0 and 4. 0 corresponds 
to the catalytic site, as found from the crystal structure. 
1..4 corresponds to each of the potential binding sites. These
different potential binding sites can also be found by the
provided software, by extensive sampling around the protein 
structure. From the initial 10,000 samples, 20 samples
that have the lowest energy are picked, and a random 
energy descent at these conformations is performed. The resulting
conformations are filtered based on their closeness to the
protein structure, as well as their pairwise distances. 
(if there are multiple closeby nodes of low energy, only one of
them is kept). The top 4 conformations in this process are 
provided in 1ldm_potBindingSiteIndices.txt. 

In order to compute the escape time from the potential (or catalytic)
binding site, computeEscapeTimesWrapper script can be used in matlab. 
This computes escape times and writes the result to a file called
escapeTimes.txt.

To compute escape time, in matlab, 
go to the directory where the files 
"transitionProbs.txt" and "close2PotBindingSite#0.txt" reside,
and then run:

computeEscapeTimesWrap(1); 

Make sure computeEscapeTimesWrap.m is either in the current
directory, or in the path. This can be done with a command
such as: 

addpath('mFiles/');

Have fun!

References:
[1] http://www.accelrys.com/insight/DelPhi_page.html
[2] Stochastic Roadmap Simulation: An efficient Representation
 and Algorithm for Analyzing Molecular Motion
M.S. Apaydin, D.L. Brutlag, C. Guestrin, D. Hsu,  J.C. Latombe and
 C. Varma, Journal of Computational Biology, (10) 257-281.
[3] A Motion Planning Approach to Flexible Ligand Binding. 
Singh, Latombe, Brutlag, ISMB 1999.