Chromosight

Python package to detect chromatin loops (and other patterns) in Hi-C contact maps.

• Associated publication: https://www.nature.com/articles/s41467-020-19562-7
• Documentation and analyses examples: https://chromosight.readthedocs.io
• scripts used for the analysis presented in the article https://github.com/koszullab/chromosightanalysesscripts

Installation

Stable version with pip:

sh pip3 install --user chromosight Stable version with conda: sh conda install -c bioconda -c conda-forge chromosight

or, if you want to get the latest development version:

pip3 install --user -e git+https://github.com/koszullab/chromosight.git@master#egg=chromosight

Usage

The two main subcommands of chromosight are detect and quantify. For more advanced use, there are two additional subcomands: generate-config and list-kernels. To get the list and description of those subcommands, you can always run:

bash chromosight --help Pattern detection is done using the detect subcommand. The quantify subcommand is used to compute pattern matching scores for a list of 2D coordinates on a Hi-C matrix. The generate-config subcommand is used to create a new type of pattern that can then be fed to detect using the --custom-kernel option. The list-kernels command is used to view informations about the available patterns.

Get started

To get a first look at a chromosight run, you can run chromosight test, which will download a test dataset from the github repository and run chromosight detect on it. You can then have a look at the output files generated.

Important options

When running chromosight detect, there are a handful parameters which are especially important:

• --min-dist: Minimum genomic distance from which to detect patterns. For loops, this means the smallest loop size accepted (i.e. distance between the two anchors).
• --max-dist: Maximum genomic distance from which to detect patterns. Increasing also increases runtime and memory use.
• --pearson: Detection threshold. Decrease to allow a greater number of pattern detected (with potentially more false positives). Setting a very low value may actually reduce the number of detected patterns. This is due to the algorithm which might merge neighbouring patterns.
• --perc-zero: Proportion of zero pixels allowed in a window for detection. If you have low coverage, increasing this value may improve results.

Example

To detect all chromosome loops with sizes between 2kb and 200kb using 8 parallel threads: bash chromosight detect --threads 8 --min-dist 20000 --max-dist 200000 hic_data.cool output_prefix

Options

```

Pattern exploration and detection

Explore and detect patterns (loops, borders, centromeres, etc.) in Hi-C contact maps with pattern matching.

Usage: chromosight detect [--kernel-config=FILE] [--pattern=loops] [--pearson=auto] [--win-size=auto] [--iterations=auto] [--win-fmt={json,npy}] [--norm={auto,raw,force}] [--subsample=no] [--inter] [--tsvd] [--smooth-trend] [--n-mads=5] [--min-dist=0] [--max-dist=auto] [--no-plotting] [--min-separation=auto] [--dump=DIR] [--threads=1] [--perc-zero=auto] [--perc-undetected=auto] chromosight generate-config [--preset loops] [--click contactmap] [--norm={auto,raw,norm}] [--win-size=auto] [--n-mads=5] [--threads=1] chromosight quantify [--inter] [--pattern=loops] [--subsample=no] [--win-fmt=json] [--kernel-config=FILE] [--norm={auto,raw,norm}] [--threads=1] [--n-mads=5] [--win-size=auto] [--perc-undetected=auto] [--perc-zero=auto] [--no-plotting] [--tsvd] <contactmap> chromosight list-kernels [--long] [--mat] [--name=kernel_name] chromosight test

detect:
    performs pattern detection on a Hi-C contact map via template matching
generate-config:
    Generate pre-filled config files to use for detect and quantify.
    A config consists of a JSON file describing parameters for the
    analysis and path pointing to kernel matrices files. Those matrices
    files are tsv files with numeric values as kernel to use for
    convolution.
quantify:
    Given a list of pairs of positions and a contact map, computes the
    correlation coefficients between those positions and the kernel of the
    selected pattern.
list-kernels:
    Prints information about available kernels.
test:
    Download example data and run loop detection on it.

```

Input

Input Hi-C contact maps should be in cool format. The cool format is an efficient and compact format for Hi-C data based on HDF5. It is maintained by the Mirny lab and documented here: https://open2c.github.io/cooler/

Most other Hi-C data formats (hic, homer, hic-pro), can be converted to cool using hicexplorer's hicConvertFormat or hic2cool. Bedgraph2 format can be converted directly using cooler with the command cooler load -f bg2 <chrom.sizes>:<binsize> in.bg2.gz out.cool. For more informations, see the cooler documentation

For chromosight quantify, the bed2d file is a text file with at least 6 tab-separated columns containing pairs of coordinates. The first 6 columns should be chrom start end chrom start end and have no header. Alternatively, the output text file generated by chromosight detect is also accepted. Instructions to generate a bed2d file from a bed file are given in the documentation.

Output

Three files are generated by chromosight's detect and quantify commands. Their filenames are determined by the value of the <prefix> argument: * prefix.tsv: List of genomic coordinates, bin ids and correlation scores for the pattern identified * prefix.json: JSON file containing the windows (of the same size as the kernel used) around the patterns from pattern.txt * prefix.pdf: Plot showing the pileup (average) window of all detected patterns. Plot generation can be disabled using the --no-plotting option.

Alternatively, one can set the --win-fmt=npy option to dump windows into a npy file instead of JSON. This format can easily be loaded into a 3D array using numpy's np.load function.

Note: the p-values and q-values provided in prefix.tsv should not be used as a criterion for filtering and are only useful for ranking calls. Their values are obtained from a Pearson correlation test and could be biased due to the dependence between contact values in the window.

Contributing

All contributions are welcome. We use the numpy standard for docstrings when documenting functions.

The code formatting standard we use is black, with --line-length=79 to follow PEP8 recommendations. We use nose2 as our testing framework. Ideally, new functions should have associated unit tests, placed in the tests folder.

To test the code, you can run:

bash nose2 -s tests/

FAQ

Questions from previous users are available in the github issues. You can open a new issue for your question if it is not already covered.

Citation

When using Chromosight in you research, please cite the pubication: https://www.nature.com/articles/s41467-020-19562-7