Statistical models for detecting differential chromatin interactions mediated by a protein.

Area of Science:

• Genomics
• Molecular Biology
• Computational Biology

Background:

• Protein-mediated chromatin interactions vary across cell types and developmental stages.
• These variations are linked to cellular development and disease states.
• Accurate detection of differential chromatin interactions is scientifically important.

Purpose of the Study:

• To propose novel statistical models for identifying differential chromatin interactions.
• To analyze two-sample Chromatin Interaction Analysis by Paired-End Tag Sequencing (ChIA-PET) count data.
• To provide computational tools for detecting protein-mediated chromatin interaction changes.

Main Methods:

• Development of a One-Step and a Two-Step statistical model.
• Utilizing ChIA-PET count data with one biological replicate per sample.
• Incorporating data dependency and locus-pair relatedness for accurate identification.

Main Results:

• Both models successfully identified differentially interacted chromatins in simulations.
• Simulations showed good agreement between model classifications and true interactions.
• The One-Step model offers efficient data utilization, while the Two-Step model provides an alternative.

Conclusions:

• The proposed models effectively detect differential chromatin interactions from ChIA-PET data.
• The developed R package MDM facilitates the application of these methods.
• This work aids in understanding the role of chromatin interactions in biological processes and diseases.