HiC_tools: a curated map of the Hi-C analysis toolbox

A survey of software, not a pipeline

The README title says it plainly: Hi-C data analysis tools and papers. The repository is a collection, and the collection is organised by task rather than by language, framework or vendor.

The ordering rule is stated at the top. Tools are added by publication date, newest on top, and unpublished tools are listed at the end of each section. That single convention does a lot of work, because it means a section read from the bottom up tells you the history of how the field approached that problem, and a section read from the top tells you what to try first.

The audience is genomicists choosing a tool, and secondarily anyone writing a methods section or teaching the field. The repository points at two sibling projects for background, Hi-C data notes and single-cell Hi-C notes, and at MDnotes for other data science and genomics notes, which tells you this is one entry in a wider set of curated references rather than a standalone product.

The licence is MIT, and the README asks for contributions through CONTRIBUTING.md, with a badge welcoming pull requests. The repository listing shows the shape of the project: README.md as the main artefact, img/ for images, CONTRIBUTING.md, LICENSE, and two CSV files named pipelines_list.csv and pipeline_comparison.csv, which turn out to be the data behind the dashboard.

The dashboard is the one thing you run

There is exactly one executable component in this repository, and it is a web app. The README gives the live address at hicatalog.streamlit.app and points at a dashboard section near the top of the page.

The files behind it are dashboard.py, run_dashboard.sh and a .streamlit/ directory, which is the conventional layout for a Streamlit application with its configuration. The dependency list is correspondingly small. The requirements file contains two entries:

streamlit>=1.32.0
pandas>=2.0.0

That tells you the app reads tabular data with pandas and renders with Streamlit, which fits the two CSV files in the repository root. pipelines_list.csv is the catalogue itself and pipeline_comparison.csv is the structured comparison table, and the app is almost certainly a filterable view over them. The comparison table has a published counterpart: the README links a detailed comparison of 13 Hi-C processing tools in Genome Biology, which is the primary source for the processing-pipeline section.

The practical value of the dashboard over the README is exactly the value of a table over a wall of text. When you want a tool that does Capture-C peaks, or one that does SNP-oriented analysis, or one that does haplotype phasing, the README gives you a heading and a list; the app gives you the same list with the filters that a newcomer actually needs.

What the app is not is an analysis environment. It does not read your data, and it does not install the tools it indexes.

The sections are a task taxonomy of the field

The table of contents in the README is auto-generated by doctoc, which tells you the list is kept in sync with the headings automatically, and it also gives the clearest view of how the field divides up.

The pipeline section covers QC and quality control, Capture-C with its peaks, HiChIP and 4C. Then come resolution improvement with simulation, normalization with a CNV-aware normalization subsection, and reproducibility as a section of its own. The compartment and structure sections follow: AB compartments, peak and loop callers, differential analysis, TAD callers, TAD benchmarking, architectural features, and differential timecourse TAD analysis.

The later sections are where the field is least settled. Machine and deep learning has its own section, then SNP-oriented tools, CNV and structural variant detection, visualization, de novo genome scaffolding, 3D modeling, deconvolution and haplotype phasing.

A papers section is layered on top of the tools, with methodological reviews and general reviews, and a technology subsection that breaks out multi-omics, Micro-C, multi-way interactions and imaging, plus separate normalization and spectral clustering entries. Courses and labs are listed as well.

That structure is the real content of the repository. A person who knows Hi-C already knows these task names. A person who does not will find that the taxonomy itself, with its separation of resolution improvement from normalization and of benchmarking from calling, is a map of the field's debates.

What the pipeline entries actually say

The summaries are short but information-dense, and reading two of them tells you the level of detail on offer.

Microket is described as a 3D genomics data preprocessing pipeline for Hi-C and Micro-C, using a read-stitch strategy that stitches complementary reads to improve mapping efficiency, removing adapters with Ktrim, low-quality cycles and PCR duplicates, and using STAR for alignment. The README reports that it outperforms Juicer, HiC-Pro, HiCUP, Fan-C and Distiller in detecting more pairs, especially at shorter distances, as well as in running time and memory, and that it was tested on simulated sim3C data and the Rao2014 dataset. Output formats are pairs, hic and cool. It is a Linux command-line tool, and the entry carries the Communications Biology paper with a DOI.

FAN-C is a Python pipeline that takes raw FASTQ, aligned with BWA or Bowtie2 and artifact filtered, or pre-aligned BAMs, and offers KR or ICE normalization. Its analysis and visualization cover contact distance decay, A/B compartment detection, TAD and loop detection, average TAD and loop profiles, saddle plots, triangular heatmaps and comparison of two heatmaps. It can run in automatic or modular mode and is compatible with .cool and .hic formats. The entry cites the Genome Biology paper.

The rest follow the same pattern with their own specifics. Galaxy HiCExplorer 3 is a web server for Hi-C, Capture Hi-C and scHi-C analysis, quality control and visualization. scHiCExplorer is a set of command-line tools for single-cell Hi-C. Cooltools is the modular Python suite for processed Hi-C data in cooler format, offering ICE normalization including trans-chromosomal, distance decay curves, A/B compartment analysis with saddle plots, and TAD and loop identification with the insulation score and HiCCUPS, working at up to 100bp resolution and with Micro-C data; the README notes it supersedes hiclib. Quaich is a snakemake pipeline for post-processing built on cooltools, chromosight, mustache and coolpuppy.

Every entry names a publication, so the citations are one click away.

The tables are the part that stays useful

A list of links decays. The CSV files are the part of this repository that can be maintained properly, and they are worth understanding.

pipeline_comparison.csv holds the comparison that the README also cites in published form: 13 Hi-C processing tools compared in detail in a Genome Biology table. For a preprocessing decision, that is the highest-value artefact in the repository, because the question is not which tool is best but which trade-off suits your data and your compute.

pipelines_list.csv is the catalogue data. The dashboard reads these files rather than parsing the Markdown, which is why the README can be regenerated or edited freely without breaking the app.

The interesting pattern is that the most durable content in a survey like this is rarely the prose summary. It is the structured row: tool, data type, normalisation method, output format, command line or web interface, publication year. That row is what a lab can copy into a methods table, and it is what survives a tool being renamed or a paper being superseded.

One thing to check is how the maintenance of those rows is done. The README explains the ordering convention and the contribution process, but does not describe a review cadence for the entries, so a tool that has not been updated in a year may still be presented the same way as one released last week.

What a catalogue cannot do for you

The value of HiC_tools is inversely proportional to how much you already know, and it stops well short of being a working environment.

First, there is no installation path. The repository installs nothing from the tools it describes. You get a link to each project's own repository, and every environment, dependency set, container and reference genome index is that project's problem. That is a deliberate division of labour, and it is the right one for a survey, but it means the list cannot answer the question a practitioner actually has, which is which combination of tools will run together on my data.

Second, there is no version information. The summaries mention publication years and papers, not versions, and the repository has no GitHub releases of its own. The tools it lists move at their own pace, and a summary written when a tool was at version 3 may describe behaviour that has since changed.

Third, the licences of the listed tools are not the MIT licence of this repository. That licence covers the README and the tables. Each project carries its own terms, and some of the software in this space is distributed under academic or non-commercial terms, so any project you pick needs its own licence check.

Finally, a list cannot be a reproduction. The reproducibility section of the catalogue points you at the relevant literature, which is the right place to look, but nothing here pins an environment for you.

The honest way to use this repository is as the first pass. Read the section for your task, shortlist two or three tools, then read those tools own papers and repositories, which is exactly the order the catalogue's structure encourages.

Editorial conclusion

HiC_tools fits a researcher choosing among Hi-C tools, writing a methods section, or orienting a newcomer to the field, and it fits a lab that wants a shortlist maintained by someone else. It is the wrong tool for anyone who needs a reproducible analysis environment, since it installs nothing for you, pins no versions and carries no environment for the tools it lists, so a pipeline choice still has to be made and installed elsewhere. Before relying on it, check the linked tools rather than the summaries, confirm the repository entries pipelines_list.csv and pipeline_comparison.csv still match the current releases of the projects they describe, and note that the last push was on 2026-10-03 with no tagged releases and an MIT licence that covers the list rather than the listed software.

Frequently asked questions

Official sources

1. Issues
2. License: MIT
3. mdozmorov/HiC_tools on GitHub
4. README