Research work · 2026
Machine Learning in Natural Product-based Drug Discovery
Gargee Mahajan, Neha Kadam, Prashant S. Kharkar
Machine Learning and Deep Learning in Drug Design
Abstract
Natural products (NPs) have been used as therapeutic agents for centuries due to their unique structure and medicinal properties. However, they also come with numerous challenges, including a complex structure, limited availability, and high-throughput screening constraints. Also, traditional approaches for NP screening and development are time-consuming, resource-intensive, and have a complex chemical space. Machine learning (ML) is emerging as an excellent tool to overcome these challenges. ML techniques, including supervised and unsupervised learning, have been employed for predicting biological activities, identifying active scaffolds, optimizing lead compounds, and guiding de novo synthesis of molecules. These molecules were trained using data from various databases. The more the data, the more efficient the ML model, particularly its prediction power. The combination of ML and NP-based drug discovery will accelerate the drug discovery process encompassing target identification, lead identification and optimization, preclinical and clinical phases, and will also reduce the time and cost in the discovery pipeline. The synergy between ML and NP research is reshaping the future of drug discovery towards a more predictive and efficient paradigm.