alma-classifier 0.1.4

A Python package for applying pre-trained epigenomic classification models

ALMA Classifier

A Python package for epigenomic diagnosis and prognosis of acute myeloid leukemia.

Models

1. ALMA Subtype: Classifies 28 subtypes (27 WHO 2022 acute leukemia subtypes + normal control)
2. AML Epigenomic Risk: Predicts 5-year mortality probability for AML patients
3. 38CpG AML Signature: Risk stratification using targeted 38 CpG panel

Installation

Docker (Recommended)

docker pull fmarchi/alma-classifier:0.1.4

pip

python -m venv .venv && source .venv/bin/activate
pip install pacmap==0.7.0
# MacOS users need `brew install lightgbm`
pip install alma-classifier
python -m alma_classifier.download_models

Usage

Docker

# Demo
docker run --rm -v $(pwd)/output:/output fmarchi/alma-classifier:0.1.4 --demo --output /output/demo_predictions.xlsx

# Your data
## Transfer your input data to ./data/
docker run --rm -v "$(pwd)/data":/data -v "$(pwd)/output":/output   fmarchi/alma-classifier:0.1.4 --input /data/your_file.pkl --output /output/your_results.xlsx

Command Line

alma-classifier --input data.pkl --output predictions.xlsx
alma-classifier --demo --output demo_results.csv
alma-classifier --input data.pkl --output results.xlsx --confidence 0.5

Input Formats

For Illumina Methylation450k or EPIC, prepare a .pkl dataset in python3.8 with the following structure:

• Rows: Samples
• Columns: CpG sites
• Values: Beta values (0-1)

For nanopore WGS, follow the standard bedMethyl format with these key columns:

• Column 1: chrom - Chromosome name
• Column 2: start_position - 0-based start position
• Column 4: modified_base_code - Single letter code for modified base
• Column 11: fraction_modified - Percentage of methylation (0-100)

Output

Results include subtype classification, risk prediction, and confidence scores. Predictions below confidence threshold (default 0.5) are marked "Not confident".

Limitations

The diagnostic model does not recognize: AML with Down Syndrome, juvenile myelomonocytic leukemia, transient abnormal myelopoiesis, low-risk MDS, or lymphomas.

Citation

Francisco Marchi, Marieke Landwehr, Ann-Kathrin Schade et al. Long-read epigenomic diagnosis and prognosis of Acute Myeloid Leukemia, 12 December 2024, PREPRINT (Version 1) available at Research Square [https://doi.org/10.21203/rs.3.rs-5450972/v1]