Beyond Bioinformatics – Revisiting Linear Interaction Energy (LIE) for Virtual Screening in Drug Design
The end-goal of bioinformatics-based approach per se and in conjunction with experiments is the identification and validation of potential therapeutic target(s). The next challenging step is to find potential ligand molecules that can hit the target. Often large-scale docking calculations are performed to virtually screen large compound libraries. Accurate prediction of the binding affinity or the binding free energy of drug-like molecules to therapeutic targets is central to computational drug design. Scoring functions are rapid in calculations but lack accuracy while rigorous free energy calculations are more accurate but computationally expensive and require expert intervention. Intermediate and simpler approach like Linear Interaction Energy (LIE) provides a balanced solution but need empirical parameterization for a given class of compound and/or therapeutic target, restricting its use in large-scale virtual screening. We hypothesized that a universal scaling parameters may exist for the LIE method, that can be applied to any set/class of ligand molecules and therapeutic target, making it amenable to virtual screening. We performed LIE calculations on >20 diverse datasets, including some benchmark datasets and across multiple projects using Gromos force field for proteins and ligands. It was observed that our hypothesis is likely to be ‘True’ and the details of the work will be presented.
Bio
Dr. Alpesh K. Malde is a Research and Teaching Academic at Griffith University and leads the Computational Chemistry & Drug Design Group. His research focuses on development, validation and applications of computational methods, algorithms and parameters and encompasses molecular dynamic simulations and free energy calculations. Dr. Malde conceptualized and co-developed the Automated Topology Builder (ATB, >25K registered user, commercially available) webserver and developed GroLigFF webserver to facilitate computational drug design. He has published over 70 peer-reviewed papers with an h-index of 28 with >6,000 citations.
