Background:The recent development of metagenomic sequencing makes it possible to massively sequence microbial genomes including viral genomes without the need for laboratory culture.Existing reference-based and gene h...Background:The recent development of metagenomic sequencing makes it possible to massively sequence microbial genomes including viral genomes without the need for laboratory culture.Existing reference-based and gene homology-based methods are not efficient in identifying unknown viruses or short viral sequences from metagenomic data.Methods:Here we developed a reference-free and alignment-free machine learning method,DeepVirFinder,for identifying viral sequences in metagenomic data using deep learning.Results'.Trained based on sequences from viral RefSeq discovered before May 2015,and evaluated on those discovered after that date,DeepVirFinder outperformed the state-of-the-art method VirFinder at all contig lengths,achieving AUROC 0.93,0.95,0.97,and 0.98 for 300,500,1000,and 3000 bp sequences respectively.Enlarging the training data with additional millions of purified viral sequences from metavirome samples further improved the accuracy for identifying virus groups that are under-represented.Applying DeepVirFinder to real human gut metagenomic samples,we identified 51,138 viral sequences belonging to 175 bins in patients with colorectal carcinoma(CRC).Ten bins were found associated with the cancer status,suggesting viruses may play important roles in CRC.Conclusions:Powered by deep learning and high throughput sequencing metagenomic data,DeepVirFinder significantly improved the accuracy of viral identification and will assist the study of viruses in the era of metagenomics.展开更多
基金The research was supported by the U.S.National Institutes of Health R01GM120624,National Science Foundation DMS-1518001,National Natural Science Foundation of China(11701546)the Simons Collaboration on Computational Biogeochemical Modeling of Marine Ecosystems(CBIOMES+1 种基金grant ID 549943)We thank Drs.Michael S.Waterman,Gesine Reinert,Ying Wang,Rui Jiang,Yang Lu,Lizzie Dorfrnan,Mr.Weili Wang,and Mr.Luigi Manna for helpful discussions and suggestions.We thank USC Center for High Performance Computing(HPC)for helping us use their cluster computers.
文摘Background:The recent development of metagenomic sequencing makes it possible to massively sequence microbial genomes including viral genomes without the need for laboratory culture.Existing reference-based and gene homology-based methods are not efficient in identifying unknown viruses or short viral sequences from metagenomic data.Methods:Here we developed a reference-free and alignment-free machine learning method,DeepVirFinder,for identifying viral sequences in metagenomic data using deep learning.Results'.Trained based on sequences from viral RefSeq discovered before May 2015,and evaluated on those discovered after that date,DeepVirFinder outperformed the state-of-the-art method VirFinder at all contig lengths,achieving AUROC 0.93,0.95,0.97,and 0.98 for 300,500,1000,and 3000 bp sequences respectively.Enlarging the training data with additional millions of purified viral sequences from metavirome samples further improved the accuracy for identifying virus groups that are under-represented.Applying DeepVirFinder to real human gut metagenomic samples,we identified 51,138 viral sequences belonging to 175 bins in patients with colorectal carcinoma(CRC).Ten bins were found associated with the cancer status,suggesting viruses may play important roles in CRC.Conclusions:Powered by deep learning and high throughput sequencing metagenomic data,DeepVirFinder significantly improved the accuracy of viral identification and will assist the study of viruses in the era of metagenomics.
