Accurate chromatin marks peak calling with Omnipeak.

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• Chromatin immunoprecipitation sequencing (ChIP-seq) is vital for mapping genomic elements.
• Existing peak callers struggle with variable peak lengths, data quality, and missing controls.
• These limitations hinder comparative and meta-analyses of ChIP-seq data.

Purpose of the Study:

• To develop a universal, unsupervised peak-calling algorithm for diverse ChIP-seq and ATAC-seq data.
• To address limitations of current tools in handling variable peak lengths and data quality.
• To provide a robust solution for comparative and meta-analyses.

Main Methods:

• Developed Omnipeak, a novel algorithm using a constrained three-state hidden Markov model.
• Modeled global genomic read coverage to capture structure patterns across all scales.
• Benchmarked Omnipeak against eight existing methods using over 850 public and synthetic datasets (ChIP-seq, ATAC-seq).

Main Results:

• Omnipeak demonstrated consistent peak calling across narrow, broad, and variable mark lengths.
• Achieved superior agreement between replicates and robustness against noise.
• Showed resilience to missing control tracks, a common issue in meta-analyses.

Conclusions:

• Omnipeak offers a universal and unsupervised approach to peak calling for ChIP-seq and ATAC-seq.
• Its ability to handle diverse data types and quality makes it ideal for large-scale analyses.
• Omnipeak enhances the reliability and consistency of genomic element mapping.