An evolutionary profile guided greedy parallel replica excahnge Monte Carlo search algorithm for rapid convergence in protein design

PLEASE DOWNLOAD AND UNZIP ProDeMo.tar.gz IN THE CURRENT DIRECTORY.

This folder contains the codes and the input files necessary for protein design.

Software requirement:
	a.	TMalign- You can get the TMalign executable from- https://zhanglab.ccmb.med.umich.edu/TM-align/
	b.	DSSP- You can get the DSSP executable from- https://swift.cmbi.umcn.nl/gv/dssp/DSSP_5.html. We have used version 2.0.4 of the same.		
	c.	SCWRL4- You can get the SCWRL4 executable from- http://dunbrack.fccc.edu/scwrl4/SCWRL4.php
	d.	FoldX- You can get the FoldX executable from- http://foldxsuite.crg.eu/academic-license-info

Download and install them in your system if they are not already available. Make sure that the programs are running standalone and can be integrated with the given code.

Provide the absolute path of the softwares and relevant folders in parameters.txt. An example file is provided as parameters_example.txt. 
Please note that parameters_example.txt is prepared in our system. Your system specific location should be in parameters.txt.
DO NO ALTER THE SEQEUNCE AND THE FORMAT OF THE parameters.txt file

Unzip XrayCullPdbDb99 and provide the absolute path of the folder in the parameters.txt file corresponding to the culledDB parameter. 

Steps to run the protein design module-

1. Place the location of the target PDB file in the parameters.txt file corresponding to the pdbPath parameter.
Ensure the following in the input PDB files-
a. Place only the ATOM records corresponding to a single chain ID.
b. There should be no chain break (backbone discontinuity) in the PDB files. You can fix it using ITASSER or MODELLER softwares.
c. The residues shouldn't have multiple occupancies.
d. The atom and residue index should be contiguous.

2. Set the sweeps parameter in the parameters.txt file. The sweeps paramter corresponds to MAXSTEP (for details see manuscript) for each replica. We have run our simualtions
with MAXSTEP=30000.

3. DO NOT CHANGE THE RELATIVE LOCATION of the INPUT, OUTPUT and bin folders. DO NOT REMOVE any file/executable from bin.

The INPUT folder contain atom, meanVarianceDb, MSA, seq, SSSAPhipsi and TMScore subfodlers. These subfolders contain intermediate files required for execution.
These files can be resused for multiple runs on the same target file. An INPUT_example directory is provided for convenience.

The OUTPUT folder will contain the file Designed_INPUTPDB.txt. Where, INPUTPDB is the name of the input pdb file. An OUTPUT_example directory is provided for convenience.

The bin folder contains the files and the the executables necessary for the execution of the protein design program.

4. Once you have set the necessary parameters, you can run the protein design program in 2 ways-
a. Run ./ProDeMo or
b. Run bsub <submitJob.sh

We have performed our simualtions utilizing 60 cores using OpenMP (The option- #BSUB -n 60 is set in submitJob.sh) corresponding to which have used nstep_max=6 
(for details see manuscript) for our simulations. The proposed greedy approach will scale up with the number of cores available. It is advisable to use atleast 32 
cores for the execution.

The average run time of the entire greedy REMC algorithm (with a maximum of 300000 decoy sequence generation across 10 simulation trajectories) with new sequences 
generated (only sequence generated and a random number considered as evolutionary energy) in each step in a Dell PowerEdge R518 computing cluster employing 60 cores 
for the 76 proteins in the benchmark dataset is 146 seconds. Whereas the average run time for third-party software SCWRL, FoldX, and the package computing evolutionary 
energy for each step (for each decoy sequence) for the same 76 proteins are 3.40 seconds, 1.90 seconds and 1.10 seconds respectively.

For any doubts and suggestions get back to
anupambanerjee@iitkgp.ac.in / pkuntal39@iitkgp.ac.in / pralay@cse.iitkgp.ac.in