chromatin-state-inference

You can infer chromatin states from histone modification ChIP-seq data by using chromHMM to binarize BED or BAM files and learn a genome-wide state annotation model based on a user-specified genome assembly, bin size, and number of states.

ChromHMM genome segmentation requires histone modification ChIP-seq files in BED or BAM format, a target genome assembly, a bin size for data binarization, and the desired number of chromatin states to learn.

Binarization in chromatin state inference processes aligned ChIP-seq data into discrete bins to determine the presence or absence of histone marks, creating the binary input matrix needed to train the chromHMM state annotation model.

Yes, you can use BAM files directly for chromatin state annotation alongside BED files, as the workflow accepts both formats to perform binarization and model learning for genome-wide segmentation.

You specify the desired number of chromatin states based on your biological context, and the chromHMM workflow learns the model parameters to produce genome-wide state annotations reflecting that exact number of states.

Chromatin state inference outputs a chromhmm_output directory containing the binarized data and learned model directories, which together provide the genome-wide state segmentation and annotation results.