CNet

CNet identifies groups of genomic signatures from genome-wide multi-omics profiling whose combined effects are significantly associated with clinical and phenotypical outcomes in complex disease.

Key Features:

• Generalized Sequential Feedforward Method: Systematically processes heterogeneous genomic signature profiles to manage and analyze diverse multi-omics data types.
• Down-Sampling Bootstrap Strategy: Employs down-sampling bootstrap resampling to reduce noise and minimize inclusion of random hitchhiking signatures.
• Dynamic Trimming Procedure: Iteratively removes less informative genomic signatures at each step to retain the most relevant signals.
• Signature Selection for Gene Representation: Selects optimal signatures to represent specific genes to facilitate precise identification of functional elements.
• Modeling Diverse Data Types: Implements four distinct models to accommodate continuous, categorical, and censored clinical and phenotypical measurements.

Scientific Applications:

• Drug-Response Data: Identifies genomic signatures associated with drug responses to inform pharmacogenomic analyses.
• Multidimensional Cancer Genomics Data: Analyzes complex cancer genomics to uncover potential disease-causing chains involving somatic mutations and pathway activities.
• Genome-Wide Association Study (GWAS) Data: Detects significant associations between genomic signatures and phenotypic outcomes across GWAS traits.

Methodology:

CNet combines a generalized sequential feedforward method with a down-sampling bootstrap strategy and a dynamic trimming procedure and applies four models for continuous, categorical, and censored data.