CavFind: A Novel Algorithm to Detect Cavities on Protein Structure

Various essential functions in living organisms are performed by binding of proteins with other molecules (ligands). Proper detection and analysis of ligand binding locations (cavities) leads towards the success of the overall drug designing system. The main challenge towards that goal is that the problem is computationally hard. In the present study, a soft computing-based algorithm has been proposed that is capable of detecting cavities on protein’s structure. The proposed algorithm is based on Voronoi decomposition and implemented by applying the self-organizing map (SOM) clustering algorithm. The proposed algorithm is evaluated with 48 protein–ligand complexes and compared with two existing algorithms like Fpocket and ConCavity and compared with the results stored in two existing databases like BioLiP and CaviDB. The proposed algorithm is also capable of computing some descriptors of the detected cavities.