AI-ASSISTED DRUG DISCOVERY TARGETING GENETIC PATHWAYS: INTEGRATING MOLECULAR BIOLOGY AND COMPUTATIONAL CHEMISTRY
The increasing complexity of disease pathogenesis and the limitations to the traditional approaches to drug discovery suggests that it is necessary to move to more effective and integrative approaches. In this study, a genome-wide genetic pathway analysis system has been built, and an artificial intelligence (AI)-assisted framework to identify the drug targets based on a genetic pathway analysis has been developed. The used dataset was the Genomics of Drug Sensitivity in Cancer (GDSC) data that included both measures of drug response and genomic features (gene expression, copy number alterations, and methylation data). The use of machine learning models like Random Forest, XGBoost, and Artificial Neural Networks were implemented in drug sensitivity prediction. The XGBoost is the best in predictive accuracy among them. Importance of feature analysis and pathway enrichment analysis revealed that key determinants of drug response are key signalling pathways, with PI3K-Akt and MAPK pathways being the most important. Moreover, a drug-target pathway network was built to explain the intricate biological interactions and to find out possible therapeutic targets. The results showed the utility of the combination of AI with pathway-level bioinformatics analysis to not only provide predictive accuracy but also provide biological interpretability. Though the use of in vitro data is limited in the study, the study provides a scalable, robust framework of drug discovery based on AI assistance. The results can be used to further the field of precision medicine through the identification of biologically relevant targets as well as optimization of therapeutic approaches.