Automatic Cavity Identification and Decomposition into Subpockets with CAVIAR

Abstract

<div>Motivation: The detection of small molecules binding sites in proteins is central to structure based</div><div>drug design. Many tools were developed in the last 40 years, but only few of them are available</div><div>today, open-source, and suitable for the analysis of large databases or for the integration in</div><div>automatic workflows. In addition, no software can characterize subpockets solely with the</div><div>information of the protein structure, a pivotal concept in fragment-based drug design.</div><div>Results: CAVIAR is a new open source tool for protein cavity identification and rationalization.</div><div>Protein pockets are automatically detected based on the protein structure. It comprises a subcavity</div><div>segmentation algorithm that decomposes binding sites into subpockets without requiring the</div><div>presence of a ligand. The defined subpockets mimick the empirical definitions of subpockets in</div><div>medicinal chemistry projects. A tool like CAVIAR may be valuable to support chemical biology,</div><div>medicinal chemistry and ligand identification efforts. Our analysis of the entire PDB and the</div><div>PDBBind confirms that liganded cavities tend to be bigger, more hydrophobic and more complex</div><div>than apo cavities. Moreover, in line with the paradigm of fragment-based drug design, the binding</div><div>affinity scales relatively well with the number of subcavities filled by the ligand. Compounds</div><div>binding to more than three of the subcavities identified by CAVIAR are mostly in the nanomolar</div><div>or better range of affinities to their target.</div><div>Availability and implementation: Installation notes, user manual and support for CAVIAR are</div><div>available at https://jr-marchand.github.io/caviar/. The CAVIAR GUI and CAVIAR command line</div><div>tool are available on GitHub at https://github.com/jr-marchand/caviar and the package is hosted</div><div>on Anaconda cloud at https://anaconda.org/jr-marchand/caviar under a MIT license. The GitHub</div><div>repository also hosts the validation datasets.</div>