Microbiome meta-analysis and cross-disease comparison enabled by the SIAMCAT machine learning toolbox

• Published in: Genome Biology Vol. 22; no. 1; p. 93
• Main Authors: Wirbel, Jakob, Zych, Konrad, Essex, Morgan, Karcher, Nicolai, Kartal, Ece, Salazar, Guillem, Bork, Peer, Sunagawa, Shinichi, Zeller, Georg
• Format: Journal Article
• Language: English
• Published: England BioMed Central 30-03-2021; BMC
• Subjects: Algorithms; Biomarkers; Computational Biology - methods; Confounding Factors, Epidemiologic; Crohn Disease - etiology; Databases, Genetic; Datasets; Gastrointestinal Microbiome; Humans; Inflammatory bowel disease; Learning algorithms; Machine Learning; Meta-analysis; Meta-Analysis as Topic; Metagenome; Metagenomics; Metagenomics - methods; Microbiome data analysis; Microbiome-wide association studies (MWAS); Microbiomes; Microbiota; Models, Statistical; ROC Curve; Software; Statistical analysis; Statistical modeling; Taxonomy; Variables; Workflow; Statistical modeling; Microbiome data analysis; Microbiome-wide association studies (MWAS); Machine learning; Meta-analysis
• ISSN: 1474-760X; 1474-7596; 1474-760X
• DOI: 10.1186/s13059-021-02306-1

The human microbiome is increasingly mined for diagnostic and therapeutic biomarkers using machine learning (ML). However, metagenomics-specific software is scarce, and overoptimistic evaluation and limited cross-study generalization are prevailing issues. To address these, we developed SIAMCAT, a versatile R toolbox for ML-based comparative metagenomics. We demonstrate its capabilities in a meta-analysis of fecal metagenomic studies (10,803 samples). When naively transferred across studies, ML models lost accuracy and disease specificity, which could however be resolved by a novel training set augmentation strategy. This reveals some biomarkers to be disease-specific, with others shared across multiple conditions. SIAMCAT is freely available from siamcat.embl.de .