X-intNMF: a cross- and intra-omics regularized NMF framework for multi-omics integration.

Area of Science:

• Computational biology
• Bioinformatics
• Systems biology

Background:

• Multi-omics data integration is crucial for understanding complex diseases.
• Existing methods often overlook cross-omics interactions, limiting biological insight.
• There is a need for models that capture both intra- and cross-omics relationships.

Purpose of the Study:

• To propose X-intNMF, a network-regularized non-negative matrix factorization (NMF) model for multi-omics integration.
• To explicitly incorporate intra- and cross-omics feature interaction networks.
• To enhance biological interaction representation and improve integration quality.

Main Methods:

• Developed X-intNMF, a network-regularized NMF model.
• Integrated mRNA expression, microRNA expression, and DNA methylation data.
• Applied the model to predict cancer phenotypes and clinical outcomes.

Main Results:

• X-intNMF outperformed state-of-the-art methods in predicting breast cancer phenotypes and classifying lung/ovarian cancer outcomes.
• Ablation studies confirmed the importance of both cross- and intra-omics interactions.
• Survival analysis revealed strong prognostic value for overall survival and disease-free status across 25 TCGA cancer datasets.

Conclusions:

• X-intNMF effectively models multi-layered molecular interactions for improved multi-omics integration.
• The model demonstrates interpretability, robustness, and scalability.
• X-intNMF offers significant potential for advancing precision medicine and understanding complex diseases.