gdock v1.0.0 - Genetic Algorithm applied to Protein-Protein Docking

In the project we apply Genetic Algorithm to protein-protein docking. Each individual is represented by its cartesian coordinates and euler angles, and the fitness of each conformation is given by a scoring function.

The input of each simulation is two PDB files, one receptor and one ligand; plus a list of residues that are likely to play a part in the interaction. There is (and probably will never be) support for ab initio experiments.

Initial placement of the molecule is done in a way that each of the interface composed by residues that are likely to be in the interface are facing each other. From here on, the receptor is fixed in space and all geometric opertions are be done to the ligand. Each individual (=conformation) is generated by randomly assigning three floats that will describe a region in space (around the geometric center of the initial ligand positioning) and three more floats that will describe its rotation.

The fitness of each one of these individual is evaluated (=scored) and over each generation, individuals have a chance of mutation, where one of the six descriptors is randomly changed and also a chance of crossover, where individuals exchange descriptors.