IDAM

IDAM identifies disease-associated microbial biomarkers by inferring disease-associated gene modules from matched metagenomic and metatranscriptomic data.

Key Features:

• Data requirements: Accepts raw sequencing data or an expression matrix from matched metagenomic and metatranscriptomic samples.
• Gene context conservation (uber‑operons): Integrates the concept of uber-operons to leverage conserved genomic neighborhoods in grouping genes.
• Gene co-expression: Incorporates gene co-expression patterns as regulatory signals for module inference.
• Mathematical graph model: Combines uber-operon information and co-expression via a mathematical graph model to identify gene modules.
• Metadata-independence: Infers disease-associated gene modules without requiring prior phenotype metadata.
• Output: Produces gene modules consisting of subsets of genes and samples associated with specific phenotypes.
• Empirical validation and performance: Applied to datasets including inflammatory bowel disease (IBD), melanoma, type 1 diabetes mellitus, and irritable bowel syndrome (IBS), with reproducibility validated on independent IBD cohorts and reported superior performance versus other tools.

Scientific Applications:

• Microbial biomarker discovery: Identification of microbial gene modules associated with disease phenotypes from metagenomic and metatranscriptomic data.
• Inflammatory bowel disease (IBD): Inference and validation of disease-associated gene modules using public and independent IBD cohorts.
• Melanoma: Application to publicly available melanoma-associated microbiome datasets to infer disease-linked gene modules.
• Type 1 diabetes mellitus (T1DM): Application to T1DM datasets for discovery of disease-associated microbial features.
• Irritable bowel syndrome (IBS): Application to IBS datasets for identifying microbial characteristics linked to the syndrome.

Methodology:

Requires raw sequencing data or an expression matrix from matched metagenomic and metatranscriptomic samples and integrates uber-operon-based gene context conservation with gene co-expression via a mathematical graph model to infer disease-associated gene modules without using prior metadata.