Infer disease-associated microbial biomarkers based on metagenomic and metatranscriptomic data

Unveiling disease-associated microbial biomarkers (e.g., key species, genes, and pathways) is an efficient strategy for the diagnosis and therapy of diseases. However, the heterogeneity and large size of microbial data bring tremendous challenges for fundamental characteristics discovery. We present IDAM, a novel method for disease- associated biomarker identification from metagenomic and metatranscriptomic data, without requiring prior metadata. It integrates gene context conservation and regulatory mechanism through a mathematical model for maximizing the number of connected components between local-low rank submatrices of a gene expression matrix and known uber-operon structures. We applied IDAM to 813 inflammatory bowel disease-associated datasets and showed IDAM outperformed existing methods in microbial biomarker identification. In addition, the identified biomarkers successfully distinguished disease subtypes and showcased their power in discovering novel disease subtypes/states. IDAM is freely available at https://github.com/OSU-BMBL/IDAM.