SNPLS

SNPLS applies a sparse network-regularized partial least squares model to integrate large-scale pairwise gene expression data, drug response profiles, and a molecular interaction network to identify joint gene–drug co-modules relevant to cancer pharmacogenomics.

Key Features:

• Integration of Diverse Data Types: Integrates large-scale pairwise gene expression data, drug response profiles, and molecular interaction networks to capture gene–drug associations.
• Sparse Network-Regularized Approach: Implements a sparse partial least squares model with network-based penalties to incorporate molecular network structure into variable selection.
• Identification of Gene-Drug Co-modules: Identifies joint co-modules comprising groups of genes and anticancer drugs associated with specific cancer cell lines.
• Application to Large-Scale Data: Applied to 13,321 gene expression profiles and pharmacological data for 98 anticancer drugs across 641 human cancer cell lines.

Scientific Applications:

• Pharmacogenomics: Maps gene–drug associations across cancer cell lines to support pharmacogenomic analyses and interpretation of drug response variability.
• Molecular Mechanism Insights: Reveals coordinated gene–drug interactions and functional modules to inform mechanisms of drug action and potential therapeutic targets.
• Cancer Research: Supports stratification of cancer cell lines and personalized medicine approaches by linking genetic profiles to anticancer drug responses.

Methodology:

Incorporates molecular networks into a sparse partial least squares model using network-based penalties and evaluates performance via simulation-based validation comparing SNPLS with typical methods.