Multimodal CustOmics

Multimodal CustOmics: A Unified and Interpretable Multi-Task Deep Learning Framework for Multimodal Integrative Data Analysis in Oncology

Hakim Benkirane ([email protected])

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

Introduction

• Multimodal CustOmics is a novel architecture for classification and survival outcome prediction.
• Multimodal CustOmics uses relevant integration between multi-omics data and histopathology slides.
• Multimodal CustOmics is able to provide both end-to-end prediction and unsupervised latent representation.
• Multimodal CustOmics has been evaluated using multiple test cases for classification and survival using TCGA datasets.
• Multimodal CustOmics is able to output multiple explainability metrics at different levels of the biological system.

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 WSI,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 WSI,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 WSI,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{benkirane2024multimodal,
  title={Multimodal CustOmics: A Unified and Interpretable Multi-Task Deep Learning Framework for Multimodal Integrative Data Analysis in Oncology},
  author={Benkirane, Hakim and Vakalopoulou, Maria and Planchard, David and Adam, Julien and Olaussen, Ken and Michiels, Stefan and Courn{\`e}de, Paul-Henry},
  journal={bioRxiv},
  pages={2024--01},
  year={2024},
  publisher={Cold Spring Harbor Laboratory}
}