Repository navigation
Releases
DMRdetector 3.2.0: Python package in one container image
DMRdetector 3.2.0 finds differentially methylated regions (DMRs) in a single patient from long-read methylation data (ONT or PacBio, hg38). It tests CpG islands and imprinted control regions against a reference atlas built from control samples.
Using the version from the manuscript? The manuscript results come from DMRdetector v2.2 (R and Nextflow): see the release v2.2-manuscript. This release is a Python rewrite with the same algorithm and parameters.
What is in this release
- A Python package with one container image (Apptainer or Docker recipe). No R, Nextflow, or conda is needed.
- Commands:
dmrdetector build(reference from controls),dmrdetector run(one sample per call),config, andget-genes. All thresholds are in onesettings.toml. - Installation with pip (
pyproject.toml, Python 3.12) as an alternative to the container. - A report on 21 disease-linked hg38 regions (imprinting disorders, uniparental disomy loci, DMPK), with direction and per-haplotype tracks.
- Inputs: ONT modkit bedMethyl, PacBio pb-CpG-tools, or a plain 5-column BED, plus a VCF for the heterozygous-SNP filter.
Install
There is no prebuilt image to download. Build it yourself from env/dmrdetector.def (Apptainer) or env/Dockerfile (Docker), or use the pip route. The commands are in the README.
Checks for this release
- Output files are identical in content to those of 3.1.0 on the demo and the end-to-end test sets, with 36 and 24 threads and with 8 and 16 threads.
- Unit tests: 438 passed with compiled kernels, 420 with
DMRDETECTOR_NO_JIT=1, and 438 withDMRDETECTOR_NO_LIBDEFLATE=1. Smoke test: 96 of 96 checks. - Stage-by-stage comparison with v2.2: doc/validation.md.
Limits
- hg38/GRCh38 only. Linux x86_64 and WSL2 only.
- The container image gives the same data files on every x86-64 CPU that we could test. A pip installation does not give this guarantee.
- Python 3.12 only;
pyproject.tomlexcludes 3.13 until it has been tested. - The Docker recipe was not built in our tests; the Apptainer recipe was.
Citation
Ou J-H*, Tsai M-JM*, Hsu R-H, Chen H-A, Wu C-S, Chao M-Y, Hsu C, Chien Y-H, Lee N-C, Hwu W-L. Detecting differentially methylated regions at CpG islands and imprinted loci with an ONT methylation atlas.
Full list of changes: CHANGELOG.md.
DMRdetector v2.2 (version used in the manuscript)
This release is the version of DMRdetector used for the manuscript:
Ou J-H*, Tsai M-JM*, Hsu R-H, Chen H-A, Wu C-S, Chao M-Y, Hsu C, Chien Y-H, Lee N-C, Hwu W-L. Detecting differentially methylated regions at CpG islands and imprinted loci with an ONT methylation atlas. <journal, year, DOI, or "in press">
What this release is
- DMRdetector v2.2 (R and Nextflow pipeline).
- All results, figures, and tables of the manuscript were produced with this code.
What the analysis used
- Reference atlas: 69 whole-blood control samples (ONT, hg38), available at https://doi.org/10.5281/zenodo.18978867
- Gene annotation: Ensembl release 86 (GRCh38).
- Runtime: the container and R environment described in the README of this tag.
Which version should I use?
- To reproduce the manuscript, use this release.
- For new analyses, use the latest version on the
mainbranch. It is a Python rewrite with the same algorithm and parameters, packaged in one container image. The comparison with v2.2 is indoc/validation.mdof that version.
Notes:
- Citation line: fill in the journal details once they are known, or leave "manuscript in press" until then. Update it after publication by editing the release notes. Editing notes does not change the tag.
- "Latest version on main": this sentence is true only after you push 3.2.0. If you publish this release first, as planned, the sentence will be correct a few minutes later. If you prefer, drop that bullet and add it when 3.2.0 is out.
- Number check: I took "69 controls" and "Ensembl release 86" from the README and doc/validation.md. Please confirm they match the final manuscript text before you publish.