Optimisation-based modelling for explainable lead discovery in malaria.

Area of Science:

• Medicinal Chemistry
• Computational Chemistry
• Drug Discovery

Background:

• Antimalarial drug resistance necessitates the urgent development of new treatments.
• The Open Source Malaria (OSM) project has screened numerous compounds, providing a foundation for further drug discovery.
• Exploring the chemical space around existing antimalarial compounds can lead to innovative therapeutic agents.

Purpose of the Study:

• To develop an explainable quantitative structure-activity relationship (QSAR) model for predicting antimalarial compound activity.
• To identify novel antimalarial lead compounds through an optimization-based QSAR methodology.
• To demonstrate the utility of explainable AI in fragment-based drug discovery.

Main Methods:

• An optimization-based quantitative structure-activity relationship (QSAR) modeling approach using piecewise regression.
• Mathematical programming formulation for explainable ligand activity modeling.
• Experimental evaluation of identified lead compounds using Plasmodium falciparum asexual growth inhibition assay (PfGIA) and human cell cytotoxicity assay.

Main Results:

• The QSAR model generated interpretable rules reflecting the contribution of chemical fragments to antimalarial activity.
• Fragment prioritization and screening of compound libraries identified potential antimalarial lead compounds.
• Three compounds were experimentally validated as potential antimalarial leads.

Conclusions:

• Explainable predictive models based on mathematical optimization can significantly enhance fragment-based lead discovery for antimalarials.
• The developed methodology offers an efficient pathway to identify and validate novel antimalarial drug candidates.
• This research contributes to the ongoing efforts to combat malaria through innovative drug discovery approaches.