Characterization Of Bio-relevance And Primary Study On Its Application In Drug Discovery

The purpose of drug screening is to search for ligands which can be combined with the target bio-macromolecules and produce biological activity, but currently existing indices (drug-likeness, lead-likeness etc.) were not considered the protein active-site information. For protein active-site information was encoded in ligands combined with protein, so ligands information can be used to guide drug design. Therefore, a novel index termed bio-relevance (BR) is proposed in this paper.BR is a measurement of biological importance and the possibility of biological origination for a compound or a database. According to Chemoinformatics methods, with application of computer programming, the characterization of BR was realized, also the potential application of this index in drug design and discovery was preliminarily studied.Firstly, based on the similarity of candidate compounds library and BRCD (Molprint 2D), a novel algorithm was designed. Through calculated of databases(DNP, CMC, MDDR, ACD-3D) proved out the rationality of BR. In addition, by exploring the same bioactivity compounds in CMC and MDDR, the compounds in different drug discovery phases and the drugs in different launched decades, we found the BR did well in distinguish drugs and non-drugs, and enhance compounds BR contributed to improve the successful rate of drug screening. This result suggests that BR can be effectively used in drug design fields.Secondly, Pri-BRCD and NPRCD sample database was respectively constructed, and the BR calculation algorithm were used to repeate the research on the above-mentioned systems, we found the discriminant power of NPRCD and pri-BRCD is much weaker than that of BRCD. This results showed that BRCD is more suitably used in drug discovery, also confirmed the accuracy of the starting point of preliminary study.In addition, Bayesian categorization algorithm was used to repeate the research on the above-mentioned systems, the advantage and disadvantage of molecular similarity algorithm and Bayesian discriminant analysis was compared. The results showed that the molecular similarity algorithm was uniform and better distinguished; also Pipeline Pilot can be used to achieve the automatic calculation of two algorithms above.Finally, through molecular descriptor calculations and correlation analysis, the influence factors of BR was discriminated. And through contrasted the relationship with Bio-relevance, toxicity and drug-likeness, the essential of BR was explored.BR is a novel drug screening index, and it was based on biological ligands. Compared with other methods, no information of negative library is needed. Also, the index value is uniform and symmetric. This index can be effectively used in some fields like the construction of compounds library, directing the synthesis of biological relevant molecules, and it has great guiding significance for drug discovery in the future.