Algorithmic development for the design of ligand and un-biased decoy databases for protein-ligand docking

LAZZARO, L; RISK, M; CAVASOTTO, CLAUDIO N.

Congreso; XX Congreso de Bioingeniería; 2015

The development and benchmarking of high-throughput docking protocols requires an adequate set of ligands and decoys. The necessity of the latter is dictated by the need to avoid artificial enrichment by generating sets of decoys with similar physical properties to those of the ligands but that are chemically diverse. The first condition avoids the artificially high enrichment caused by the separation of simple physical properties, whereas the second reduces the likelihood of the decoy sets containing actual binders (which would artificially lower the enrichment). In this work, we implement algorithms to design in silico libraries of small-molecule ligands extracted from the web, and from there, to build decoy libraries ensuring a ligand-decoy similarity of six physical properties, while enforcing ligand-decoy chemical dissimilarity. Code is developed in Python, and ligand and decoy databases will be tested in virtual docking experiment.