A computationally-efficient framework for identifying potential allosteric residues at the protein surface and within the interior

The original stress server on AWS is currently down due to AWS settings udpdates. We apologize for the inconvenience

The current version of this server has a security vulnerability. This vulnerability, which is in the rack module, has been classified by GitHub as a moderate severity security vulnerability. More information can be found here: https://nvd.nist.gov/vuln/detail/CVE-2018-16471 and also here: https://nvd.nist.gov/vuln/detail/CVE-2015-3225

Python, Perl, C compiler, MMTK

Before running the code, you need to compile the C package, type:

			```
			cd INSTALLFOLDER/surface_residues/code_pack
			make clean
			make
			cd INSTALLFOLDER/interior_residues/networkTools
			make clean
			make
			```

./stress -I INPUTFILE -O OUTPUTDIR -L LOGFILE [-s|--surface] [-i|--interior] [-h|--help]
        -I | --input:       followed by the inputfile (pdb file)
        -O | --output:      followed by the output directory
        -L | --logfile:     followed by the logfile
        -s | --surface:     run the surface-critical module(optional)
        -i | --interior:    run the interior-critical module(optonal)
        -h | --help:        for help(optional)

To run the program, you have to at least choose one of surface/interior modules, or both.

For detailed description of the two modules, please see below.

Inspired by the binding leverage framework first introduced by Mitternacht and Berezovsky in 2011, this module efficiently identifies pockets on the protein surface such that their occlusion with small ligands is predicted to drastically affect conformational change – such surface sites are thus inferred to be allosteric hotspots. Monte Carlo simulations are used to probe the all-atom protein surface with flexible ligands. After automated thresholding, those sites at which the simulated ligand strongly interferes with conformational change are given as output in the form of a ranked list of sites, with each site being represented by its list of constituent residues, as given by their residue IDs and corresponding chains. Each site usually contains 10 residues, and never more than 10. In the output file provided (ex: “3D3D__SURFACE_CRITICAL_residues.dat”), each row corresponds to a particular allosteric surface site, and columns contain the following data:

Column 1:    Index of the binding site from Monte Carlo simulations (this is used by the 
             STRESS developers for the purpose of thesholding the list of ranked sites)
Column 2:    Binding leverage score for site – higher values signify that the site is 
             such that occlusion by ligand strongly interferes with large-scale protein 
             motions, and thus likely has a stronger allosteric affect
Columns 3+:  The list of surface-critical residues in the given site, in the format 
             <RESID_CHAIN>

1. If you have any difficulty generating any of the output files, output files for this example 
are stored in the directory ./output_files/
2. When submitting a PDB file, we suggest using the biological assembly file (not the raw 
asymmetric unit).

An earlier version of this code was developed by Anurag Sethi and John Eargle of the Luthey-Schulten Group at The University of Illinois at Urbana-Champaign (see information in code headers). Extensive modifications have been introduduced by Shantao Li and Declan Clarke of the Gerstein Lab at Yale University. A few of the most notable modifications now integrated into the new version include:

1. The printing of resID and chain information in output files
2. The use of anisotropic network models that are generated by MMTK
3. Modified parameters of the attractive potentials for ligand-protein interactions
4. For the purpose of optimizing efficiency: Local searching supported by hashing is used in each 
   sampling step in the search for surface-critical residues.
5. For the purpose of optimizing efficiency: revised data structure, workflow and cut back on some 
   expensive operations such as square root.

Allosteric residues within the protein interior often function as essential communication ‘bottlenecks’ between distinct regions. To identify such bottleneck residues, this computationally efficient module represents a protein structure as a network of interacting residues, and simultaneously incorporates information about conformational change. The list of residues given as output are thus predicted to function in essential communication bottlenecks within the protein, and are thus potentially allosteric (see our publication for details). The output file provided (ex: “3D3D__INTERIOR_CRITICAL_residues.txt”) has the following format:

Column 1:   RESIDUE (the 0-based index of the residue in the processed PDB file)
Column 2:   ResID (as it appears in the PDB file) 
Column 3:   Residue type (ex: “VAL”)
Column 4:   Chain

1. If you have any difficulty generating any of the output files, output files for this example 
are stored in the directory ./output_files/

2. When submitting a PDB file, we suggest using the biological assembly file (not the raw 
asymmetric unit).