hic-contact-map-feature-annotation

Annotate Features on Hi-C Contact Maps Using Juicer Command Line Tools

Summary

This skill applies Juicer's command-line post-processing tools to annotate structural features (loops, domains, peaks) on pre-generated Hi-C contact map files (.hic format). It enables quantitative extraction of 3D genome organization features for downstream analysis and interpretation.

When to use

You have completed Hi-C map generation (producing .hic files from aligned reads) and need to detect and annotate topological features such as chromatin loops, topologically associating domains (TADs), or interaction peaks. This is the standard post-processing step in the Juicer pipeline when the contact matrix is ready but features have not yet been called.

When NOT to use

• Raw FASTQ reads or alignment files are your input — use the Juicer pipeline (chimeric/merge/dedup/final stages) first to generate the .hic file.
• Feature annotations already exist in another format — this skill is for de novo feature calling, not format conversion or merging of existing annotations.
• You require sub-kilobase resolution features — Juicer is optimized for kilobase-resolution Hi-C data; ultra-high-resolution methods (e.g., Micro-C) may require specialized tools.

Inputs

• .hic contact map file (binary Hi-C matrix)
• genome identifier (e.g., hg19, mm10)
• optionally: restriction enzyme site file or custom reference genome

Outputs

• annotated loop coordinates file (e.g., loop_calls.bedpe or equivalent)
• domain boundary annotations (e.g., domain list with start/end positions)
• peak-calling results (for HiCCUPS: peak coordinates and significance scores)

How to apply

Load the .hic contact map file into Juicer's command-line tools suite, which provides specialized algorithms for feature detection (e.g., HiCCUPS for loop calling, domain detection for TAD boundaries). Select the appropriate tool based on your target feature type and resolution requirements. Execute the chosen tool with parameters tuned to your sequencing depth and resolution; for GPU-accelerated peak calling (HiCCUPS), CUDA support is required, but CPU alternatives exist. The tool outputs annotated feature coordinates (loop anchors, domain boundaries) in standard formats. Validation occurs by visual inspection in Juicebox and by checking that detected features align with expected biological patterns (e.g., loops at known CTCF sites, domain sizes in expected ranges).