De novo virus inference and host prediction from metagenome using CRISPR spacers

Abstract

Viruses are the most numerous biological entity, existing in all environments and infecting all cellular organisms. Compared with cellular life, the evolution and origin of viruses are poorly understood; viruses are enormously diverse and most lack sequence similarity to cellular genes. To uncover viral sequences without relying on either reference viral sequences from databases or marker genes known to characterize specific viral taxa, we developed a virus inference method based on clustered regularly interspaced short palindromic repeats (CRISPR). CRISPR is a prokaryotic nucleic acid restriction system that stores memory of previous exposure. Our method can reconstruct viral sequences targeted by CRISPR and predict their hosts using unassembled short-read metagenomic sequencing data. Analysing human gut metagenomic data, we extracted 11,391 terminally redundant CRISPR-targeted sequences which are likely complete circular genomes of viruses or plasmids. The sequences include 257 complete crAssphage family genomes, 11 genomes larger than 200 kilobases, 766 genomes of Microviridae species, 114 genomes of Inoviridae species and many entirely novel genomes of unknown taxa. We predicted the host(s) of approximately 70% of discovered genomes by linking protospacers to taxonomically assigned CRISPR direct repeats. These results support that our method is efficient for de novo inference of viral genomes and host prediction. In addition, we investigated the origin of the diversity-generating retroelement (DGR) locus of the crAssphage family. Phylogenetic analysis and gene locus comparisons indicate that DGR is orthologous in human gut crAssphages and shares a common ancestor with baboon-derived crAssphage; however, the locus has likely been lost in multiple lineages recently.

UG/Abi: a highly diverse family of prokaryotic reverse transcriptases associated with defense functions

“…Next, we tried to infer whether these putative phages were predicted to be active or not by comparing them with databases of CRISPR spacers derived from human gut metagenomes. To this end, we used a publicly available collection of spacers extracted from 11, 817 human gut metagenome datasets (64) and an in-house spacer database that was built by running CRISPRCasFinder (65) on the Integrative Human Microbiome Project – Inflammatory Bowel Disease metagenomic dataset (66) (Methods). By doing this, we obtained that 1346 out of 1447 dereplicated viral genomes harboring UG27 (93, 02%) were targeted by at least 1 spacer, with 1197 (82%) being targeted by 5 or more spacers, suggesting a recent active role of these viral genomes in their natural environment.…”

“…Host taxonomy prediction of UG27 viral genomes was done by performing CRISPR spacers comparisons against a reference database using CrisprOpenDB (63). As most of the taxonomy predictions pointed out towards bacteria present in the gut microbiome, viral genomes were compared against recent human gut metagenomic spacers database (64) and a custom collection of spacers found in CRISPR arrays from the human gut microbiome; compiled by running CRISPRCasFinder (65) on all metagenomic contigs from the HMP2-IBD database (66). For the comparison, BLASTn (67) with -task “blastn-short” option was used, and only hits with ≤2 mismatches and >95% sequence identity were kept.…”