A34 Automated profiling of the human virome from raw metagenomic data

Abstract

Viruses influence human health as conventional pathogens, asmodulators of gene expression and through their involvementin complex host-microbiome interactions. Next generationsequencing (NGS) has enabled us to explore the role of themicrobiome in human health and disease. Metagenomicsequencing should allow us to profile all biological elements ina clinical sample in an unbiased, hypothesis-free way. Howeverviruses display much greater variation than all other elements,and the existing tools and methods for virus identification and discovery are not effective enough. We have developed a bioinformatics pipeline to identify and classify all known virusespresent in a metagenomic sample. Viral NGS reads are identified using a protein-based alignment method, DIAMOND, whichis substantially faster than the standard BLAST method, andmore reliable for viruses. These reads are automaticallyassembled into contigs using SPAdes, a de novo assembler. Thecontigs are then used to classify the virus at species level usinga pan-viral typing tool based on all available taxonomic reference sequences from the International Committee onTaxonomy of Viruses (ICTV) database. This bioinformatics pipeline is Java-encoded and will include an easy-to-use web interface that is fit-for-purpose for researchers or clinicians. Thistool can assemble viral contigs from paired-end reads generatedby an Illumina MiSeq sequencer. So far 1865 viruses can be identified at species level resolution and 10 viruses (chikungunyavirus, dengue virus, HBV, HCV, HHV8, HIV-1, HPV, HTLV-1, YFV,and Zika virus) at the genotype level. A web version of the panviral typing tool is already available and a web version withextended NGS functionality is currently being evaluated.Eliminating the need for virus-specific laboratory techniques, ortargeted sequence capture, means a virome can be profiled inthe context of its non-viral microbiome. Preliminary findingssuggest our tool offers greater functionality than existing alternatives, with greater sensitivity to known viruses (includingbacteriophages), automatic assembly and good quality phylogenetic analyses. A systematic comparison is underway.