HOCOMOCO: a comprehensive collection of human transcription factor binding sites models

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Transcription factor binding site (TFBS) models are essential for understanding gene regulation.
• Existing repositories often contain multiple, technique-specific TFBS models for a single transcription factor (TF).
• A unified TFBS model per TF is more practical for computational applications.

Purpose of the Study:

• To develop a comprehensive collection of high-quality TFBS models for human TFs.
• To improve TFBS model accuracy by integrating diverse experimental data.
• To provide a pragmatic resource for transcription regulatory network reconstruction.

Main Methods:

• Integration of TFBS data from low- and high-throughput experimental methods.
• Construction of position weight matrices using ChIPMunk software, including a novel periodic positional prior mode.
• Systematic curation and quality assessment of TFBS models, selecting one model per TF where possible.

Main Results:

• The HOCOMOCO database was created, featuring 426 curated TFBS models for 401 human TFs.
• 172 models in HOCOMOCO are derived from multiple data sources, demonstrating successful data integration.
• Integrated models generally show improved quality, likely due to reduced technique-specific bias.

Conclusions:

• Integrating diverse experimental data yields more robust and accurate TFBS models.
• HOCOMOCO provides a valuable, curated resource for studying human transcription regulation.
• The approach enhances the reliability of computational reconstruction of regulatory networks.