DeepToA: an ensemble deep-learning approach to predicting the theater of activity of a microbiome

• DOI: 10.1093/bioinformatics/btac584
• Publication Date: 2022-08-27T12:58:32Z
• Authors: Wenhuan Zeng; Anupam Gautam; Daniel H Huson
• Keywords: Taxonomic rank; Biological classification

Abstract Motivation Metagenomics is the study of microbiomes using DNA sequencing. A microbiome consists an assemblage microbes that associated with a ‘theater activity’ (ToA). An important question is, to what degree does taxonomic and functional content former depend on (details the) latter? Here, we investigate related technical question: Given and/or profile estimated from metagenomic sequencing data, how predict ToA? We present deep-learning approach this question. use both profiles as input. apply node2vec embed hierarchical into numerical vectors. then perform dimension reduction clustering, address sparseness data thus make problem more amenable algorithms. Functional features are combined textual descriptions protein families or domains. ensemble framework DeepToA for predicting ToA amicrobial community, based profiles. SHAP (SHapley Additive exPlanations) values determine which prediction. Results Based 7560 downloaded MGnify, classified 10 different theaters activity, demonstrate has accuracy 98.30%. show adding information increases accuracy. Availability implementation Our available at http://ab.inf.uni-tuebingen.de/software/deeptoa. Supplementary Bioinformatics online.