INDEED: R package for network based differential expression analysis.

Area of Science:

• Computational biology
• Bioinformatics
• Genomics and transcriptomics

Background:

• Omics technologies have advanced, increasing data availability for identifying disease-associated biomolecules.
• Conventional differential expression analysis methods focus on mean and variance, potentially missing complex molecular interactions.
• Network-based approaches consider biomolecular interactions but can generate overly complex networks using correlation methods.

Purpose of the Study:

• To introduce INDEED, an R/Bioconductor package for constructing sparse networks using partial correlation.
• To identify biomolecules with significant changes at both individual expression and pairwise interaction levels.
• To enhance disease biomarker discovery by integrating network information.

Main Methods:

• Development of the INDEED R/Bioconductor package.
• Construction of sparse networks utilizing partial correlation.
• Identification of biomolecules based on differential expression and interaction patterns.
• Application to cancer biomarker discovery datasets.

Main Results:

• INDEED successfully constructed sparse networks from omics data.
• The package identified disease-associated biomolecules by considering both expression and interaction changes.
• Analysis of cancer datasets demonstrated INDEED's utility in ranking potential biomarkers.

Conclusions:

• INDEED offers a novel computational method for disease-associated biomolecule identification.
• Biomolecules selected by INDEED show potential for improved sensitivity and specificity in disease detection.
• The INDEED framework can be extended for multi-omic data integration to discover reliable biomarkers.