DRIM: A Web-Based System for Investigating Drug Response at the Molecular Level by Condition-Specific Multi-Omics...

Area of Science:

• Pharmacogenomics and computational biology.
• Multi-omics data analysis and systems biology.

Background:

• Understanding gene-drug interactions is crucial for drug discovery and personalized medicine.
• Existing methods struggle to analyze diverse transcriptome and multi-omics data, especially time-series data, due to complex gene interactions.
• Integrating multi-omics data and time-series analysis is challenging but essential for interpreting drug responses.

Purpose of the Study:

• To develop an integrative framework, DRIM, for analyzing multi-omics and time-series data to understand drug responses.
• To identify perturbed sub-pathways and regulatory mechanisms influenced by drug treatments.
• To provide a tool for analyzing drug effects at the molecular level.

Main Methods:

• Developed Drug Response analysis Integrating Multi-omics and time-series data (DRIM) framework.
• Utilized tensor decomposition and autoencoder deep learning for multi-omics data integration to identify potential mediator genes.
• Employed influence maximization to construct regulatory networks involving transcription factors, mediator genes, and perturbed pathways.

Main Results:

• Identified potential mediator genes associated with drug sensitivity (IC50, AUC) using multi-omics data.
• Constructed time-varying perturbed sub-pathways to illustrate dynamic drug effects.
• Built a regulatory network to map influence paths from transcription factors to perturbed pathways.

Conclusions:

• DRIM effectively analyzes complex multi-omics and time-series data to reveal drug-induced pathway perturbations.
• The framework aids in understanding sub-pathway regulatory mechanisms and drug sensitivity in cancer cell lines.
• DRIM offers a valuable tool for advancing pharmacogenomics research and personalized medicine.