metilene³: Identifying DMRs Across Multiple Conditions with Auto-Classification

metilene³ is a computational tool to identify Differentially Methylated Regions (DMRs) across multiple groups (supervised) or samples (unsupervised). With the identified DMRs, metilene³ enables inference of epigenetic relationships by constructing a Differentially Methylated Tree (DMTree), which can also be used for sample clustering.

Please see the metilene³-doc for more details.

Installation

You can download and install metilene³ on Linux, WSL and macOS from this GitHub repo:

git clone https://github.com/zzhu1372/metilene3.git
cd ./metilene3
make

Dependencies can be installed with conda:

conda create -y -n metilene3 -c bioconda -c conda-forge python==3.10.0 pandas pandarallel scikit-learn seaborn biopython gseapy r-base bioconductor-ChIPseeker bioconductor-org.Hs.eg.db bioconductor-txdb.hsapiens.ucsc.hg19.knowngene bioconductor-txdb.hsapiens.ucsc.hg38.knowngene
conda activate metilene3

Please check here for more details.

Quick Start

After installation, you can test metilene³ with the included test dataset demo_input.tsv:

python ./metilene3.py -test True -o demo_output

You should get these files in the demo_output folder:

DMRs-unsupervised.tsv
DMRs.tsv
clusters.tsv
group-ID.tsv
report.html

It should take less than one minute on your laptop.

If success, you can run metilene³ with your methylation dataset (see format):

python ./metilene3.py -i your_methylation.tsv -o your_output

Check the full tutorial to customize your command.