TransPilot: mining key transcription factors by correlating binding sites with differentially expressed genes

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Existing software for identifying key transcription factors (TFs) regulating differentially expressed genes (DEGs) has limitations.
• Accurate identification of regulatory TFs is crucial for understanding gene expression and cellular processes.

Purpose of the Study:

• To present TransPilot, a novel web server for identifying key transcription factors from transcriptome data.
• To overcome limitations in current methods for mining regulatory TFs and DEGs.

Main Methods:

• TransPilot employs weighted Kendall's tau rank correlation to compare TF-target sets with DEG lists.
• It utilizes an artificial neural network (ANN) trained on log-likelihood ratio (LLR) models to identify transcription factor binding sites (TFBSs).
• An imputation method addresses missing TFBS data, and expression profile correlations determine regulatory direction (positive/negative).

Main Results:

• TransPilot successfully identifies key TFs by correlating ranked TF-target sets with DEG lists, emphasizing high-weight genes.
• The server was benchmarked using transcriptome data from a macrophage polarization experiment, demonstrating its efficacy.
• The method effectively ranks TF-target pairs based on binding potential and regulatory direction.

Conclusions:

• TransPilot provides an effective and user-friendly web server for identifying key transcription factors from transcriptome data.
• The tool enhances the analysis of gene regulation and provides valuable insights into biological pathways.
• The methodology offers a robust approach for TF-DEG association studies.