MLOmics: Cancer Multi-Omics Database for Machine Learning.

Area of Science:

• Bioinformatics
• Computational Biology
• Cancer Research
• Machine Learning

Background:

• Machine learning (ML) shows promise in multi-omics cancer research.
• High-quality, preprocessed datasets are crucial for ML model development.
• Existing public databases like TCGA and LinkedOmics are not optimized for ML models.

Purpose of the Study:

• Introduce MLOmics, an open cancer multi-omics database.
• Facilitate the development and evaluation of bioinformatics and ML models.
• Provide a readily usable resource for cancer multi-omics analysis.

Main Methods:

• Developed MLOmics, an open-access database.
• Integrated multi-omics data from 8,314 patient samples across 32 cancer types.
• Included stratified features, extensive baselines, and support for downstream analysis.

Main Results:

• MLOmics offers comprehensive multi-omics data for all 32 cancer types.
• The database contains 8,314 patient samples with four omics types.
• Features include stratified data, baselines, and bio-knowledge linking.

Conclusions:

• MLOmics serves as a valuable resource for ML-driven cancer research.
• The database streamlines the development and evaluation of bioinformatics tools.
• Facilitates interdisciplinary analysis in cancer multi-omics.