Analysis and Classification of Human Microbiomes: Detection of Bioindicators and Optimization through Machine Learning

• Jonas Coelho Kasmanas USP / Helmholtz Centre for Environmental Research–UFZ / University of Leipzig
• Ulisses Nunes da Rocha Helmholtz Centre for Environmental Research–UFZ
• Peter F Stadler University of Leipzig
• André Carlos Ponce de Leon Ferreira de Carvalho USP

DOI: https://doi.org/10.5753/sbcas_estendido.2026.20228

Resumo

The human microbiome plays a crucial role in health and disease, yet extracting meaningful insights from it remains challenging. This work presents an integrated framework: (1) the HumanMetagenomeDB, standardizing metadata from 69,000 human metagenomes; (2) MuDoGeR, a multi-domain genome recovery tool; (3) an evaluation of genome recovery biases; (4) gSpreadComp, a workflow for bacterial resistance-virulence; and (5) an AutoML-driven platform built on prokaryotic genomes enabling biologically interpretable feature selection. Together, these contributions enable researchers to move from raw metagenomic data to actionable biological insights at scale.

Referências

Kasmanas, J. C., Bartholomäus, A., Corrêa, F. B., Tal, T., Jehmlich, N., Herberth, G., von Bergen, M., Stadler, P. F., Carvalho, A. C. P. d. L. F. d., and Nunes da Rocha, U. (2021). HumanMetagenomeDB: a public repository of curated and standardized metadata for human metagenomes. Nucleic Acids Research, 49(D1):D743–D750.

Kasmanas, J. C., Magnúsdóttir, S., Zhang, J., Smalla, K., Schloter, M., Stadler, P. F., de Leon Ferreira de Carvalho, A. C. P., and Rocha, U. (2025). Integrating comparative genomics and risk classification by assessing virulence, antimicrobial resistance, and plasmid spread in microbial communities with gspreadcomp. GigaScience, 14:giaf072.

Pasolli, E., Asnicar, F., Manara, S., Zolfo, M., Karcher, N., Armanini, F., Beghini, F., Manghi, P., Tett, A., Ghensi, P., Collado, M. C., Rice, B. L., DuLong, C., Morgan, X. C., Golden, C. D., Quince, C., Huttenhower, C., and Segata, N. (2019). Extensive Unexplored Human Microbiome Diversity Revealed by Over 150,000 Genomes from Metagenomes Spanning Age, Geography, and Lifestyle. Cell, 176(3):649–662.

Rocha, U., Kasmanas, J. C., Kallies, R., Saraiva, J. P., Toscan, R. B., Štefanič, P., Bicalho, M. F., Borim Correa, F., Baştürk, M. N., Fousekis, E., Viana Barbosa, L. M., Plewka, J., Probst, A. J., Baldrian, P., and Stadler, P. F. (2024a). MuDoGeR: Multi-Domain Genome recovery from metagenomes made easy. Molecular Ecology Resources, 24(2):e13904.

Rocha, U., Kasmanas, J. C., Toscan, R., Sanches, D. S., Magnusdottir, S., and Saraiva, J. P. (2024b). Simulation of 69 microbial communities indicates sequencing depth and false positives are major drivers of bias in prokaryotic metagenome-assembled genome recovery. PLOS Computational Biology, 20(10):e1012530.

Thomas, A. M. and Segata, N. (2019). Multiple levels of the unknown in microbiome research. BMC Biology, 17(1):48.

Wang, C., Wu, Q., Weimer, M., and Zhu, E. (2019). FLAML: A Fast and Lightweight AutoML Library. Version Number: 3.