A successful drug may not be the most potent ligand against its therapeutic target, but it must have a balanced profile of efficacy, selectivity, PK, toxicity, etc. It is both time consuming and cost ineffective to scale up drug candidates and experimentally determine the related properties. An ideal scenario is to instantly profile a drug candidate; a better scenario is to receive the profile even before the compound is synthesized; while the best scenario is to visualize the impacts of in situ modifications of a drug molecule on the profile of absorption, PK, and toxicity. The molecular diagnostic system enables users to compare profiles of drug candidates, to analyze the changes of the profile a drug candidate upon structural modification to facilitate the decision-making process in de novo drug design. The current system covers physicochemical properties (logP, aqueous solubility, PSA), CYP450 liabilities (CYP1A2, CYP2C19, CYP2C9, CYP2D6, and CYP3A4), PK (PAMPA, P-gp, and RLM), and toxicity (hERG, PLD, and AMES). The QSAR predictive models are mostly trained by a few thousand structurally diverse compounds to achieve outstanding robustness and predictive power. The atom-type-based molecular descriptors grant the models with excellent interpretability. The hot spots of each molecule are highlighted for different properties, providing users clear clues to modify the structure for an improved profile. The Molecular Diagnostic System boasts an easy-to-use interface, and a neat design to visualize and analyze the molecular profiles. It is a powerful tool for medicinal chemists and computational chemists to carry out daily based rational drug design.

Molecular diagnostic system: A step toward virtual pharmaceutical company

251st National Meeting (2016)

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