CNet

CNet is a software tool that tackles the challenge of identifying functional genes associated with or causing phenotypic outcomes. It leverages the wealth of data generated from genome-wide multi-omics profiling of complex diseases. This innovative tool is designed to discover associations between various genomic signatures and clinical or phenotypical outcomes by identifying groups of genomic signatures whose combinatory effects are significantly associated with these outcomes.

At the core of CNet is a generalized sequential feedforward method, which is enhanced by a down-sampling bootstrap strategy aimed at reducing random hitchhiking signatures. This method ensures that the analysis focuses on the most relevant genomic signatures by dynamically trimming less informative ones at every process step. One of the key strengths of CNet is its ability to manage heterogeneous genomic signature profiles simultaneously, allowing it to select the best signature to represent a specific gene accurately.

CNet is adaptable, introducing four models to accommodate continuous, categorical, and censored data. This flexibility makes CNet suitable for various biological and clinical applications, including drug-response analysis, multidimensional cancer genomics, and genome-wide association studies for multiple traits. Testing across these varied scenarios has demonstrated CNet's effectiveness in identifying signatures associated with outcomes, and its application has extended to identifying likely disease-causing chains involving somatic mutations, pathway activities, and patient outcomes.