Abductive reasoning applied to modeling chemical biology can only be made possible if both quality datasets (i.e. in vitro, HTS or expert curation) and versatile cheminformatics and molecular modeling methods are available. Identifying compounds that are susceptible to metabolic degradation via cytochrome P450 (CYP450) pathways, and identifying which major isoform is responsible is an ongoing problem. Here we demonstrate using a combination of 2D and 3D cheminformatics and modeling tools built into MOE (Molecular Operating Environment) how a rational workflow can be developed to predict CYP450 isoform specificity. Using a combination of carefully curated in vitro CYP450 data and a variety of modeling techniques implemented in MOE (binary classification trees, 2D-molecular fingerprints, pharmacophores) we developed efficient computational approaches to identify putative CYP interacting ligands. These models are easily further integrated into a KNIME workflow, using the MOE extensions for KNIME. The tiered molecular triage workflow we developed using MOE can be used to interrogate large datasets of compounds for potential CYP450 binders with sub-type profiling capabilities. These modeling approaches can be directly generalized to other protein families (i.e. nuclear receptors or the kinome), in which isoform specificity and poly-pharmacological interactions are of interest. These studies place emphasis on the need for accurate curated data in addition to versatile and inter-operable modeling platforms, such as MOE, that enable multi-dimensional modeling approaches for interrogating chemical biology for large scale molecular profiling.

Multi-target molecular profiling using MOE: A CYP450 isoform selectivity case study

256th National Meeting (2018)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.