Factor-specific generative pattern from large-scale drug-induced gene expression profile.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Drug discovery necessitates identifying disease-specific drug targets.
• Understanding drug-induced gene expression is crucial for target identification.

Purpose of the Study:

• To introduce FacPat, a novel computational approach for identifying optimal factor-specific patterns in drug-induced gene expression.
• To reveal context-specific gene effects and novel relationships between drugs, diseases, and genes.

Main Methods:

• Utilized a genetic algorithm based on pattern distance to mine factor-specific patterns from the LINCS L1000 dataset.
• Applied Benjamini-Hochberg correction to control the false discovery rate.
• Performed functional enrichment analysis to characterize biological features.

Main Results:

• Identified significant and interpretable factor-specific patterns involving 480 genes, 7 chemical compounds, and 38 human cell lines.
• Discovered genes exhibiting context-specific effects influenced by chemical compounds and/or cell lines.
• Revealed novel associations between chemical compounds, cell lines, and gene expression profiles.

Conclusions:

• FacPat effectively identifies optimal factor-specific patterns explaining drug-induced gene expression.
• The approach facilitates the discovery of context-specific gene effects, crucial for understanding drug mechanisms.
• FacPat demonstrates utility in uncovering novel relationships critical for advancing drug discovery and development.