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

Abstract

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 revealed some biomarkers to be disease-specific, others shared across multiple conditions. SIAMCAT is freely available from siamcat.embl.de.