multiDEGGs: Single or Multiomic Differential Network Analysis for Biomarker Discovery and Feature Engineering for...

Area of Science:

• Bioinformatics
• Computational Biology
• Systems Biology

Background:

• Clinical trials increasingly use high-throughput omic data for patient stratification and biomarker discovery.
• Traditional differential gene expression analysis has limitations in interpreting networked molecular data.
• Differential network analysis offers a complementary approach for comparative studies.

Purpose of the Study:

• To present multiDEGGs, a CRAN R package for differential network analysis in single or multiomic settings.
• To enable interactive exploration of cross-omic patterns and interactions.
• To facilitate feature selection and identification of biologically relevant interactions for machine learning pipelines.

Main Methods:

• Utilizes a multiomic graph framework to generate and evaluate differential networks for each data type.
• Integrates networks into a comprehensive visualization for exploring cross-omic patterns.
• Facilitates integration into cross-validation machine learning pipelines for feature selection.

Main Results:

• Validated using rheumatoid arthritis patient cohorts, identifying multiomic differential interactions.
• Eight machine learning models were trained to predict treatment resistance using RNA sequencing data.
• multiDEGGs showed an average improvement of 0.10 in AUC compared to conventional feature selection methods.

Conclusions:

• multiDEGGs provides a robust framework for multi-omic differential network analysis.
• The package enhances biomarker discovery and patient stratification in clinical trials.
• It offers improved performance in machine learning applications for predicting treatment outcomes.