3D shape-based analysis of cell line-specific compound response in cancers.

The rapid increase in the volume of high-throughput anticancer chemical screening data requires a better interpretation of the relationships between diverse chemical structures and their varied effects in distinct cancer subtypes. Unexpected compound efficacy or resistance in cancer cells has been difficult to explain, in part because there has been no systematic analysis of compound response profiles in cancer cells with different genotypic backgrounds. In this study, we compared 2D chemical- and 3D shape-based similarity search methods to study the structure-activity relationships of anticancer compounds in a collection of heterogeneous cancer cell lines. The 3D shape-based metric provided better resolution than the 2D chemical topology-based method for identifying compound pairs with similar cellular response profiles. We confirmed that the 3D method exclusively identified compound pairs with different chemical scaffolds that stimulated highly similar cellular responses. The present analyses provide useful guidelines for investigating the lineage- and genotype-specific activities of diverse compounds and their mechanisms of action.