Motif-guided sparse decomposition of gene expression data for regulatory module identification

Area of Science:

• Computational Biology
• Genomics
• Systems Biology

Background:

• Genes function in coordinated modules and networks.
• Gene clustering identifies co-expressed genes but often fails to capture co-regulated genes.
• Accurate identification of gene regulatory modules is crucial for understanding biological processes.

Purpose of the Study:

• To develop a novel computational approach for identifying gene regulatory modules.
• To integrate gene expression data with DNA sequence motif information for improved module discovery.
• To enhance the accuracy of identifying co-regulated genes beyond simple co-expression.

Main Methods:

• Developed motif-guided sparse decomposition (mSD), a two-step algorithm.
• Step 1: Motif-guided clustering to estimate transcription factor activity.
• Step 2: Sparse component analysis to estimate gene regulation strength and predict target genes.

Main Results:

• mSD successfully identified functionally distinct gene modules in yeast data, enriched with validated transcription factors.
• Applied to breast cancer cell line data, mSD uncovered key gene regulatory modules relevant to endocrine therapy.
• Demonstrated improved performance in finding regulatory modules compared to traditional methods.

Conclusions:

• Motif-guided sparse decomposition (mSD) offers an integrated strategy for regulatory module identification.
• mSD balances co-regulation and co-expression for accurate transcription factor activity estimation.
• The mSD method provides context-specific regulatory modules applicable to diverse biological studies.