CustOmics

CustOmics: A versatile deep-learning based strategy for multi-omics integration

Hakim Benkirane ([email protected])

Oncostat Team, U1018 Inserm, CESP Laboratory of mathematics and informatics of CentraleSupelec

Introduction

• CustOmics is a novel architecture for classification and survival outcome prediction.
• CustOmics uses a new integration strategy for a more versatile multi-omics integration.
• CustOmics is able to provide both end-to-end prediction and unsupervised latent representation.
• CustOmics has been evaluated using multiple test cases for classification and survival using TCGA datasets.
• CustOmics is able to explain, to a certain degree, classification results.

Paper Link: Link to the published paper

Downloading TCGA Data

To download omics data (formatted as .tsv files) and other clinical metadata, please refer to the NIH Genomic Data Commons Data Portal and the cBioPortal.

Running Experiments

Experiments can be executed through the script main.py, the basic usage to run a tumor type classification on the Pancancer dataset is as follows:

python main.py --cohorts PANCAN --sources CNV,RNAseq,methyl --task classification --data_directory DATA_DIRECTORY --result_directory RESULTS_DIRECTORY

To run PAM50 classification task on TCGA-BRCA dataset:

python main.py --cohorts TCGA-BRCA --sources CNV,RNAseq,methyl --task classification --data_directory DATA_DIRECTORY --result_directory RESULTS_DIRECTORY

To run survival tasks on specific datasets:

python main.py --cohorts TCGA-BLCA,TCGA-BRCA,TCGA-LUAD,TCGA-GBM,TCGA-UCEC --sources CNV,RNAseq,methyl --task survival --data_directory DATA_DIRECTORY --result_directory RESULTS_DIRECTORY

License

This source code is licensed under the MIT license.

Cite us

Citation

If you use this code in your research, please cite our paper.

@article{benkirane2023,
    doi = {10.1371/journal.pcbi.1010921},
    author = {Benkirane, Hakim AND Pradat, Yoann AND Michiels, Stefan AND Cournède, Paul-Henry},
    journal = {PLOS Computational Biology},
    publisher = {Public Library of Science},
    title = {CustOmics: A versatile deep-learning based strategy for multi-omics integration},
    year = {2023},
    month = {03},
    volume = {19},
    url = {https://doi.org/10.1371/journal.pcbi.1010921},
    pages = {1-19},
    number = {3}
}