Next-Generation Cheminformatics Approaches for Rational Drug Discovery

D. Fourches
North Carolina State University,
United States

Keywords: cheminformatics, drug discovery, MD-QSAR, docking, machine learning


Cheminformatics is the scientific field that uses computers to solve chemical problems. The rise of artificial intelligence powered by the skyrocketing power of computers is offering new opportunities to develop and apply new strategies for the rational design of chemicals. In this presentation, we will discuss new cheminformatics methods to characterize, model, visualize, and predict chemicals’ activity towards a given biological target. We will notably present (i) the development of predictive Quantitative Structure-Activity Relationships (QSAR) using machine learning techniques and numerical descriptors computed from GPU-accelerated molecular dynamics simulations (MDS) of large series of protein-ligand complexes; (ii) the use of AI, 3D-docking and MDS to study complex HLA-drug-peptide tripartite complexes relevant for drug-induced idiosyncratic adverse events, (iii) the AI-powered enumeration of virtual compounds with the example of a 1 million library of new macrolide scaffolds, and (iv) new techniques and online services to analyze and visualize the chemical space. The aforementioned methods are of high relevance for drug discovery as well as chemical risk assessment, green chemistry, and nanomaterial design.