mixNBHMM

mixNBHMM detects and classifies genomic regions with differential enrichment across multiple conditions in multi-replicate epigenomic experiments.

Key Features:

• Three-State Hidden Markov Model (HMM): Employs a three-state HMM to identify differential broad peaks across the genome.
• Finite Mixture of Negative Binomials: Models the emission distribution in the differential state as a finite mixture of negative binomial distributions to accommodate diverse ChIP-seq signal profiles, including short and broad peaks.
• Multi-Condition and Multi-Replicate Analysis: Optimized for analysis across multiple experimental conditions and biological replicates.
• Combinatorial Pattern Classification: Classifies specific combinatorial patterns of differential epigenomic activity to characterize complex regulatory states.

Scientific Applications:

• Epigenomics Research: Applied to study interactions between the human genome and proteins to investigate regulation of cellular processes.
• Disease Association Studies: Identifies changes in epigenomic activity across conditions to explore associations with complex diseases.
• Chromatin Regulatory States Characterization: Aids in characterizing chromatin regulatory states to inform understanding of gene regulation and expression patterns.

Methodology:

Integrates read-count data from ChIP-seq, ATAC-seq, DNase-seq and related experiments and applies a three-state HMM whose differential state uses a finite mixture of negative binomial emission distributions to model epigenomic signal variability and classify differential peaks.