Sequence-based Deep Learning Reveals the Bacterial Community Diversity and Horizontal Gene Transfer

摘要

Metagenomics is the application of advanced genomic techniques to the study of microbial communities in their native environments, and does not require lab cultivation nor isolation of individual genomes. Here, we propose a novel approach for metagenomics operational taxonomic unit (OTU) assignment using deep learning. The experimental results demonstrate that a PCA-based convolutional neural network is very powerful and fast for metagenomics OTU assignment and also prioritizing horizontal gene transfer events.