Cheminformatics is a multi-disciplinary field at the interface between chemistry and data science. Cheminformatics education has been part of graduate training for students in the Division of Chemical Biology and Medicinal Chemistry at the UNC Eshelman School of Pharmacy. Typical graduate students in our Introduction to Molecular Modeling and Drug Design class have solid chemical biology background but little experience in computational molecular modeling methods. In the past few years, we have developed multiple workflows integrating KNIME and Python to facilitate student training in major cheminformatics concepts and practical applications. These concepts include but are not limited to: Application Programming Interfaces of important chemogenomics databases such as ChEMBL and PubChem to retrieve respective content; chemical and biological data curation and integration; exploratory analysis of complex chemogenomics data using multiple data analytical techniques, such as hierarchical clustering analysis, self-organizing maps, activity cliff landscape; best practices for Quantitative-Structure Activity Relationship (QSAR) modeling for continuous value and classification datasets; and virtual screening of large chemical libraries. Our experience suggests that integrating KNIME and Python allows us to facilitate the comprehension and execution of key steps in a typical cheminformatics study. In a recent class project, we engaged students in modeling multiple targets of neurotoxicity and neuroprotection. We will describe our educational workflows and the results of this class project as an example of a successful “learn-by-doing” educational strategy.

Integrating KNIME and python for teaching cheminformatics and data science

ACS Spring 2022

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