CuNA
Complex diseases are characterized by their intricate etiology, combining a variety of factors and spanning a diverse set of omics data types. To obtain insights on disease subtypes, biomarkers associated with them,and predict disease states, we need to integrate and analyze multiomics data. Here, we propose a cumulant-based network analysis toolkit (CuNA) which finds higher-order interactions between multiomics variables and identify disease subtypes, discover biomarkers, visualize them in an interactive dashboard, and provide a novel method to compute cumulant-based risk scores (CuRES) from multiomics data.
Framework
Input
CuNA takes a csv file as input with the features in columns and samples in rows (see ./docs/data/tcga_mixomics_new_train_forcuna.csv). It is multithreaded and takes an argument for number of threads.
Output
cd CuNA outputs three components:
Networks
Network file with three columns, v1,`v2`,`count`, corresponding to two vertices and the interaction/edge term between them.
A file with the communities corresponding to the features.
A file with the node rank (importance of node)
CuRES
A risk score called CuRES computed from multi-omics data.
Usage
See tutorial
Contact
`
Aritra Bose (a dot bose at ibm dot com)
`
Citation
Please cite the following article if you use CuNA: Bose, A., Platt, D. E., Haiminen, N., & Parida, L. (2021). CuNA: Cumulant-based Network Analysis of genotype-phenotype associations in Parkinson’s Disease. medRxiv, 2021-08.