A model-based approach to transcription regulatory network reconstruction from time-course gene expression data

Area of Science:

• Computational Biology
• Genomics
• Molecular Biology

Background:

• Time-course gene expression profiling reveals dynamic cellular responses.
• Transcription factors (TFs) orchestrate gene regulation during these dynamic processes.
• Experimental methods like ChIP-seq are limited in identifying all relevant TFs.

Purpose of the Study:

• To develop a computational method for inferring functional TF binding sites from time-course gene expression data.
• To identify key TFs regulating co-expressed genes in dynamic biological processes.
• To overcome limitations of experimental TF identification techniques.

Main Methods:

• A regression-based model was developed to infer TF binding sites.
• The model incorporates association strength between TF-target gene pairs.
• Lasso-penalized regression was used to identify the most informative TF-target interactions.
• Promoter sequence analysis was employed to assess binding site computational predictions.

Main Results:

• The proposed computational method successfully inferred TF-target interactions from gene expression profiles.
• The model identified biologically meaningful TF-target pairs.
• Application to E2-induced apoptosis in MCF-7 cells demonstrated the method's utility.

Conclusions:

• The developed regression-based model is effective for identifying regulatory transcription factors from gene expression dynamics.
• This computational approach provides valuable insights into molecular mechanisms underlying cellular responses.
• The method offers a complementary strategy to experimental techniques for TF identification.