If you dont already have numpy and scipy installed, it is best to download
Anaconda, a python distribution that has them included.
https://continuum.io/downloads
PyPI install, presuming you have all its requirements installed:
pip install ngsfragments
pip install MethylVerse
The developmental version of MethylVerse can also be install from GitHub.
pip install git+https://github.com/kylessmith/MethylVerse/
First use of the MethylVerse software will initialize a download of necessary reference data. Raw reference data is also available at on Zenodo.
https://zenodo.org/records/16580408
https://zenodo.org/records/16581863
General useage:
import MethylVerse as mv
beta_values = mv.core.read_methylation("path/to/methylation")
classifier = mv.tools.classifiers.MPACT.MPACT_classifier_torch.MPACT_classifier()
predictions = classifier.predict(beta_values)Run the M-PACT classifier from the cammandline
python -m MethylVerse MPACT example.bedgraph --impute --regress --call_cnvs --verbose
Additional options can be viewed:
python -m MethylVerse MPACT -h
usage: __main__.py MPACT [-h] [--impute] [--regress] [--probability_threshold PROBABILITY_THRESHOLD]
[--max_contamination_fraction MAX_CONTAMINATION_FRACTION] [--call_cnvs] [--out OUT] [--verbose]
input_data
positional arguments:
input_data Input data, idat name, nanopore bed file, or MethylDackel bedGraph
options:
-h, --help show this help message and exit
--impute Impute data missing CpGs
--regress Whether to regress data and remove background CSF/Immune
--probability_threshold PROBABILITY_THRESHOLD
Probability threshold for M-PACT classification (default: 0.7)
--max_contamination_fraction MAX_CONTAMINATION_FRACTION
Max contamination fraction for M-PACT classification (aggressiveness of removing background, default=0.3)
--call_cnvs Call CNVs from the methylation file
--out OUT Output file
--verbose Verbose output