BayesPeak

BayesPeak applies a fully Bayesian hidden Ising model to detect enriched regions (peaks) in ChIP-seq data for genome-wide identification of protein-DNA interactions, DNA methylation, and histone modifications.

Key Features:

• Bayesian hidden Ising model: Implements a fully Bayesian hidden Ising model that dynamically constructs signal profiles for each chromosome to model spatial dependencies in ChIP-seq tag distributions.
• Modeling of tag distributions: Accounts for both global and local distributions of sequence tags to characterize enrichment patterns.
• Systematic error detection: Includes model-diagnosis capabilities to identify falsely enriched regions caused by sequencing or mapping errors.
• One-sample and two-sample analyses: Supports both one-sample and two-sample experimental designs for comparative analyses.
• High sensitivity and low FDR: Reports high sensitivity and spatial resolution for transcription factor binding site detection while maintaining a lower false discovery rate compared to MACS, CisGenome, and SISSRs.

Scientific Applications:

• Protein–DNA interaction mapping: Genome-wide peak detection for transcription factors and other DNA-binding proteins from ChIP-seq data.
• Epigenomic profiling: Identification of regions associated with DNA methylation and histone modifications.
• Comparative binding analysis: Detection and comparison of differential binding or enrichment between conditions or samples using one-sample and two-sample analyses.

Methodology:

Constructs dynamic chromosome-specific signal profiles using a fully Bayesian hidden Ising model that models spatial dependencies and global/local tag distributions, includes model-diagnosis to flag sequencing or mapping artifacts, and supports one-sample and two-sample analyses.