TRRUST: a reference database of human transcriptional regulatory interactions

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• Reconstructing transcriptional regulatory networks (TRNs) is a significant challenge in human genetics.
• Accurate evaluation of computational methods for inferring gene regulatory interactions requires gold-standard datasets.
• Existing resources may not comprehensively capture human TF-target interactions.

Purpose of the Study:

• To develop and present TRRUST, a literature-curated database of human transcriptional factor (TF)-target interactions.
• To provide a reliable benchmark for assessing computational methods used in TRN reconstruction.
• To offer additional analytical features for TF and target gene analysis.

Main Methods:

• A sentence-based text-mining approach was utilized for manual curation of regulatory interactions.
• Approximately 20 million MEDLINE abstracts were processed to extract TF-target relationships.
• The TRRUST database was compiled, containing curated interactions and associated metadata.

Main Results:

• TRRUST contains 8,015 human TF-target interactions involving 748 TF genes and 1,975 non-TF genes.
• It is presented as the largest publicly available database of its kind, curated from literature.
• TF-target pairs in TRRUST showed high enrichment in top-scored interactions from high-throughput data analysis.

Conclusions:

• TRRUST serves as a valuable and reliable benchmark for the computational reconstruction of human TRNs.
• The database facilitates the validation and improvement of methods inferring gene regulatory interactions.
• TRRUST offers unique features for exploring TF regulation, cooperativity, and associated biological pathways and diseases.