Algorithms in Bioinformatics


Deep-learning prediction of the "theater of activity" of a microbiome based on taxonomic and functional profiles of metagenomic sequencing data.


DeepToA webservice is available here.

"DeepToA" is an ensemble deep-Learning approach to predicting the "theater of activity" in a microbiome. Metagenomics is the study of microbiomes using DNA sequencing. A microbiome consists of an assemblage of microbes that is associated with a ``theater of activity'' (ToA). To what degree does the taxonomic and functional content of the former depend on the (details of the) latter? More technically, given a taxonomic and/or functional profile estimated from metagenomic sequencing data, how to predict the associated ToA? 
Here we present a deep learning approach to this question. We use both taxonomic and functional profiles as input. We apply node2vec to embed hierarchical taxonomic profiles into numerical vectors. We then perform dimension reduction using clustering, to address the sparseness of the taxonomic data and thus make it more amenable to deep learning algorithms. Functional features are combined with textual descriptions of protein families or domains. We present an ensemble deep-learning framework DeepToA for predicting the `theater of activity'' of microbial community, based on taxonomic and functional profiles. To the best of our knowledge, DeepToA is the first approach to predict the ToA of the microbiome by both taxonomic and function information of the sample, as well as offering the embedding vector of every taxon included by GTDB and NCBI. 

Publication:  Wenhuan Zeng, Anupam Gautam, Daniel H Huson, DeepToA: an ensemble deep-learning approach to predicting the theater of activity of a microbiome, Bioinformatics, 2022;, btac584. 

Preprint: Wenhuan Zeng, Anupam Gautam, Daniel H. Huson, DeepToA: An Ensemble Deep-Learning Approach to Predicting the Theater of Activity of a Microbiome bioRXiv (2022)

DeepToA webservice is available here.