PredictBind 1.0.7

PredictBind is a wrapped inspired and built around P2Rank as ligand binding site predictor for proteins

PredictBind

PredictBind is a wrapper inspired and built around P2Rank.

OS: Linux or Mac

This wrapper was written on an M1 mac and tested on UNIX systems. Requires python3 and java(versions 8 to 20).

Dependency check

python3 --version
java --version

Should return something like

Python 3.x.y

and

openjdk version "11.0.22" 2024-01-16
OpenJDK Runtime Environment Homebrew (build 11.0.22+0)
OpenJDK 64-Bit Server VM Homebrew (build 11.0.22+0, mixed mode)

Installation

Make sure you have the listed dependencies and install binder with pip.

pip install PredictBind

Commands

usage: PredictBind.py [-h] [-d {direcory_name}] [-e {file}.pdb] [-f {file}.pdb] [-t {core_num}]
                 [--files [{file1}.pdb {file2}.pdb ...]] [-v] [-ch] [-pm]

Use PredictBind.py to predict ligand binding sites of proteins from their .pdb files.

options:
  -h, --help            show this help message and exit
  -d {direcory_name}, --directory {direcory_name}
                        Analyse all files located in one local directory.
  -e {file}.pdb, --evalpredict {file}.pdb
                        Path/to/pdb.
  -f {file}.pdb, --file {file}.pdb
                        Path/to/pdb.
  -t {core_num}, --threads {core_num}
                        Specify num. of working threads for parallel dataset processing
  --files [{file1}.pdb {file2}.pdb ...]
                        Analyse many local pdb files.
  -v, --verbose         Get more detailed output of the process to the standard error.
  -ch, --chimera        Open with chimera immediately when file is ready to be visualised
  -pm, --pymol          Open with pymol when file is ready to be visualised

Tutorials

TBD.

Examples

One PDB file

PredictBind.py -f 1gln.pdb 

Dir

• If you have a directory of pdb files you want to analyse, run the --directory (-d) method:

PredictBind.py -d directory_name/

Multiple files

PredictBind.py --files 1gln.pdb 2ew2.pdb subfol1/1gln.pdb  

Visualisation

The output can directly visualised with the --chimera (-ch) and --pymol (-pm) switch, given that you have them installed on your computer, by including the switch when running the comand. Note: this only woks for the single-file methods (-f); for the --directory and --local_many you need to open the visualisation cmd files manually (see: Output below).

Ex.

PredictBind.py -f 1gln.pdb -ch

Output

A prediction for a file {pdb}.pdb will create the following structure in the folder in which PredictBind.py was executed.

predict_{pdb}/
├── {pdb}.pdb_predictions.tsv
└── visualizations/
    ├── chimera_{pdb}.cmd
    ├── {pdb}.pdb.pml
    └── data/
        ├── {pdb}.pdb_points.pdb.gz
        └── {pdb}.pdb 

TSV file:

The tsv file lists the predicted pockets in order of their score. Each pocket has the following attributes:

• rank

• score

• probability

• sas_points - (int) number of solvent accessible surface points

• surf_atoms - (int) integer of the number of surface atoms

• center_x - (float) the predicted pockets x center

• center_y - (float) the predicted pockets y center

• center_z - (float) the predicted pockets z center

• residue_ids - (py dict) the residue sequence numbers that create the pocket { Chain : [ residue sequence numbers ] }

• residue_names - (py dict) the residue names that create the pocket { Chain : [ residue names ] }

• residue_types - (py dict) the character of the residues that create the pocket { Chain : [ characters ] }

  • 'N' represents non-polar amino acids
  • 'P' represents polar amino acids
  • '+' represents positively charged amino acids
  • '-' represents negatively charged amino acids
  • '0' a specific residue for which there is no info in the program

• surf_atom_ids - (py list) the atom serial number of all the atoms that are on the surface of the pocket

Information from the PDB is taken in this fashion:

Columns	Data	Justification	Data Type
1-4	"ATOM"	left	character
7-11	Atom serial number	right	integer
13-16	Atom name	left*	character
17	Alternate location indicator	-	character
18-20	Residue name	right	character
22	Chain identifier	-	character
23-26	Residue sequence number	right	integer
27	Code for insertions of residues	-	character
31-38	X orthogonal Angstrom coordinate	right	floating
39-46	Y orthogonal Angstrom coordinate	right	floating
47-54	Z orthogonal Angstrom coordinate	right	floating
55-60	Occupancy	right	floating
61-66	Temperature factor	right	floating
73-76	Segment identifier (optional)	left	character
77-78	Element symbol	right	character
79-80	Charge (optional)	-	character

Chimera CMD file

Has neccessary information to create the visualisations in chimera. The pockets are saved as selections titled "Pocket{NUM}" and colored untill the 18th pockets.

The colors are ranked the same in all output, so it can be a visual aid for quick understanding of the pockets' rankings. This is true only for the Chimera file, not for the PyMol, because in PyMol the pockets and their colors can be easily viewed in the side panel.

Color	Rank
red	1
orange	2
yellow	3
green	4
cyan	5
blue	6
medium blue	7
purple	8
hot pink	9
magenta	10
white	11
gray	12
black	13
tan	14
slate gray	15
dark khaki	16
plum	17
rosy brown	18

run :

chimera {path}/predict_{pdb}/visualizations/chimera_{pdb}.cmd

PyMol CMD file

In PyMol the pockets and their colors are conveniently displayed in the side panel. run:

pymol {path}/predict_{pdb}/visualizations/{pdb}.pdb.pml

Theoretical background

TBD

Analysis

TBD

References

This software is a lightweight version of p2rank.

• Krivak R, Hoksza D. P2Rank: machine learning based tool for rapid and accurate prediction of ligand binding sites from protein structure. Journal of Cheminformatics. 2018 Aug.
• Capra JA, Laskowski RA, Thornton JM, Singh M, Funkhouser TA. Predicting protein ligand binding sites by combining evolutionary sequence conservation and 3D structure. PLoS
• Connolly M. Solvent-accessible surfaces of proteins and nucleic acids. Science. 1983;221(4612):709–13.
• Huang B, Schroeder M. LIGSITEcsc: predicting ligand binding sites using the Connolly surface and degree of conservation. BMC Struct Biol. 2006 Sep 24;6:19.
• Krivák R, Hoksza D. Improving protein-ligand binding site prediction accuracy by classification of inner pocket points using local features. J Cheminform. 2015 Apr 1;7:12.
• Le Guilloux V, Schmidtke P, Tuffery P. Fpocket: an open source platform for ligand pocket detection. BMC Bioinformatics. 2009 Jun 2;10:168.
• RICHARDS, E M. (1977). Ann. Rev. Biophys. Bioeng. 6, 151-176.
• Zheng X, Gan L, Wang E, Wang J. Pocket-based drug design: exploring pocket space. AAPS J. 2013 Jan;15(1):228-41.