G2S: A New Deep Learning Tool for Predicting Stool Microbiome Structure From Oral Microbiome Data

• DOI: 10.3389/fgene.2021.644516
• Publication Date: 2021-04-09T07:00:50Z
• Authors: Simone Rampelli; Marco Fabbrini; Marco Candela; Elena Biagi; Patrizia Brigidi; Silvia Turroni
• Keywords: Human Microbiome Project; Gut microbiome; Oral Microbiome

Deep learning methodologies have revolutionized prediction in many fields and show the potential to do same microbial metagenomics. However, deep is still unexplored field of microbiology, with only a few software designed work microbiome data. Within meta-community theory, we foresee new perspectives for development application algorithms human microbiome. In this context, developed G2S, bioinformatic tool taxonomic fecal directly from oral data individual. The uses convolutional neural network trained on paired samples populations across globe, which allows inferring stool at family level more accurately than other available approaches. can be used retrospective studies, where sampling was not performed, especially paleomicrobiology, as unique opportunity recover related ancient gut configurations. G2S validated already characterized sample pairs, then applied dental calculi, derive putative intestinal components medieval subjects.