Microbiome meta-analysis and cross-disease comparison enabled by the SIAMCAT machine learning toolbox
Problem
Machine learning on human microbiome data lacked dedicated software and suffered from overoptimistic evaluation and poor cross-study generalization.
Why it matters
Reliable microbiome biomarkers could enable diagnosis and therapy of human diseases if models transfer across studies.
What the AI did
SIAMCAT, an R toolbox for ML-based comparative metagenomics, ran a meta-analysis of fecal metagenomic studies and used a novel training set augmentation strategy to restore cross-study accuracy.
Results
Meta-analysis of 10,803 samples; naive cross-study transfer lost accuracy and disease specificity, resolved by training set augmentation.
Impact
Provides a freely available toolbox enabling reproducible ML biomarker analysis across microbiome studies.
Record
| Venue | Genome biology |
|---|---|
| Published | 2021 |
| Work type | article |
| Authors | Jakob Wirbel, Konrad Zych, Morgan Essex, Nicolai Karcher, Ece Kartal, Guillem Salazar, Peer Bork B, Shinichi Sunagawa et al. |
| Citations | 265 (OpenAlex, at import time) |
| References | 121 |
| Model or method | SIAMCAT ML toolbox (specific algorithms not stated in source) |
| Organization | EMBL (implied by siamcat.embl.de) |
| Primary source | https://doi.org/10.1186/s13059-021-02306-1 |
| Open access | gold |
| Keywords | microbiome, machine learning, meta-analysis, metagenomics, biomarkers, R toolbox |
Importance rating
real but narrower advance
Abstract states the human microbiome is increasingly mined for diagnostic and therapeutic biomarkers; toolbox enables reproducible cross-study ML biomarker analysis on 10,803 samples.