CNet: a multi-omics approach to detecting clinically associated, combinatory genomic signatures

Area of Science:

• Genomics
• Computational Biology
• Systems Biology

Background:

• Genome-wide multi-omics profiling is crucial for understanding complex diseases.
• Identifying functional genes driving phenotypic outcomes remains a significant challenge.

Purpose of the Study:

• To develop a computational method, CNet, for identifying groups of genomic signatures associated with clinical and phenotypical outcomes.
• To address the challenge of detecting functional genes contributing to complex diseases.

Main Methods:

• CNet employs a generalized sequential feedforward approach with down-sampling bootstrap and dynamic trimming.
• It handles heterogeneous genomic data and selects optimal gene signatures.
• Four models accommodate continuous, categorical, and censored clinical data.

Main Results:

• CNet effectively identified outcome-associated signatures in drug-response, cancer genomics, and GWAS data.
• The method successfully pinpointed disease-causing chains involving mutations, pathway activities, and patient outcomes.
• Demonstrated efficacy across various biological and clinical scenarios.

Conclusions:

• CNet is a versatile tool for identifying functional gene groups linked to complex disease phenotypes.
• The method facilitates the discovery of disease mechanisms from multi-omics data.
• CNet offers a robust approach for analyzing genomic signatures and clinical outcomes.