yamet: Yet Another Methylation Entropy Tool

yamet is a C++ command line tool for computing entropies from DNA methylation data.

It takes per-cell cytosine reports, a reference file of all CpG positions, and a BED file of genomic regions, then computes sample entropy (per cell, per interval) and 2-mer Shannon entropy (across cells, per interval). Missing positions are handled via the reference file.

See also:

• https://github.com/emsonder/MethQuant
• https://github.com/emsonder/MethQuant-analysis

Note

yamet is under active development.

Installation

Brew

brew tap atchox/brew
brew install yamet

# or build from source
brew install --HEAD yamet

Compiled binaries

Compiled binaries are available on the releases page.

Build from source

git clone https://github.com/imallona/yamet.git
cd yamet/method
bash build.sh
./build/yamet --help

Usage

Usage: yamet -c <cytosine report>... -r <reference> -i <interval> [OPTIONS]

Input:
  -c,--cytosine-report,--cell           per-cell cytosine report file(s), tab-separated,
                                        sorted by chromosome and position:
                                          chr  pos  meth_reads  total_reads  rate
  --metadata                            metadata file (alternative to -c), tab-separated:
                                          cell_id  cluster  path/to/cytosine_report
  -r,--cytosine-locations,--reference   reference file of all CpG positions (required):
                                          chr  pos
  -i,--regions,--intervals              BED file of regions to compute entropies for (required):
                                          chr  start  end
  --skip-header                         lines to skip in all input files
  --skip-header-cytosine-report         lines to skip in cell files
  --skip-header-cytosine-locations      lines to skip in reference file
  --skip-header-regions                 lines to skip in intervals file
  --skip-header-metadata                lines to skip in metadata file

Output:
  -d,--det-out                          path to detailed output file
  -n,--norm-det-out                     path to normalized detailed output file
  -m,--meth-out                         path to average methylation output file
  -o,--out                              path to simple output file
  --all-meth / --templated-meth         include all CpGs in methylation summaries (default: false)

Resources:
  --cores                               cores for parallel file parsing (default: 0, auto)
  --chunk-size                          read buffer size per file, e.g. 64K, 128M (default: 64K)

Verbose:
  --print-intervals                     print parsed intervals file
  --print-reference                     print parsed reference file
  --print-sampens / --no-print-sampens  print computed sample entropies (default: true)

Misc:
  -h,--help                             show help message
  --version                             show version information

Repository layout

• .github/workflows and test: CI and build automation
• method: C++ source. See releases for binaries.
• workflow: Snakemake workflow for real datasets (CRC, Ecker, Argelaguet) and simulations.

Reproducing analyses

cd workflow
snakemake --use-conda --cores NUM_CORES

Specific targets:

# coverage simulations
snakemake --use-conda --cores NUM_CORES simulations/results/simulation_figure2.html

# within/between cell simulations
snakemake --use-conda --cores NUM_CORES simulations/results/simulation_08_combined_figure_adj.html

Simulations share the Argelaguet gastrulation download, so the tarball is fetched only once.

License

GPLv3

Contact

izaskun.mallona at gmail.com

Acknowledgements

We thank Mark D. Robinson at UZH for partly funding this project, the Swiss National Science Foundation, and the Graduate Campus at the University of Zurich for their support.

yamet builds on great open source components including snakemake and uses public data from epigenomic atlases.