VirusSeeker, a computational pipeline for virus discovery and virome composition analysis

Abstract

The advent of Next Generation Sequencing (NGS) has vastly increased our ability to discover novel viruses and to systematically define the spectrum of viruses present in a given specimen. Such studies have led to the discovery of novel viral pathogens as well as broader associations of the virome with diverse diseases including inflammatory bowel disease, severe acute malnutrition and HIV/AIDS. Critical to the success of these efforts are robust bioinformatic pipelines for rapid classification of microbial sequences. Existing computational tools are typically focused on either eukaryotic virus discovery or virome composition analysis but not both. Here we present VirusSeeker, a BLAST-based NGS data analysis pipeline designed for both purposes. VirusSeeker has been successfully applied in several previously published virome studies. Here we demonstrate the functionality of VirusSeeker in both novel virus discovery and virome composition analysis.

Intestinal virome changes precede autoimmunity in type I diabetes-susceptible children

“…In the present study, we detected a significantly lower abundance of Circoviridae-related sequences and low virome diversity in cases. The differences between these studies could be related to differences in sampling method, sequencing technique, or viral sequence detection methods (27), or to the smaller cohort of the previous study (96 samples vs. our 220 samples) (20,21). We found reduced statistical significance when we divided the case subjects into two subgroups, with and without progression to T1D.…”

“…We obtained a mean of 1.93 ± 0.92 million sequences per sample, of which a mean of 68.5% ± 14.3% were of high quality (Dataset S1). Deduplicated sequences were analyzed using VirusSeeker (27) to detect bacteriophage and eukaryotic viral sequences, which accounted for 0.04-93.17% of deduplicated sequences. There were no significant differences in the total number of sequences, the number of quality-controlled unique sequences, or the percentage of eukaryotic viral sequences between cases and controls.…”

“…We used two complementary approaches to compare intestinal viromes between cases and controls. First, we used read-based analysis to detect bacteriophage and eukaryotic viral sequences based on the best BLAST hit when queried against National Center for Biotechnology Information (NCBI) nucleotide (NT) and protein (NR) databases (27). Viral abundances were normalized to total deduplicated sequences (referred to as "relative abundance" herein) from each sample and used for statistical comparison of viromes.…”

“…Sequences from each sample were individually assembled to minimize the possibility of chimeric sequence formation (27). The taxonomic identity of each contig was assigned based on the best BLAST hit when queried against the NCBI NT and NR databases (27).…”

Next-generation Sequencing of the DNA Virome from Fecal Samples

“…Many sequence processing software tools are open-source (such as BBTools, http://jgi.doe.gov/data-and-tools/bbtools/), as are statistical analysis and graphing packages in R (https://www.r-project.org/). We used VirusSeeker (Zhao et al , 2017), a customized automated bioinformatics pipeline based on VirusHunter (Zhao et al , 2013), to detect sequences sharing nucleotide and amino acid sequence similarity to known viruses (Figure 1 below). We recommend using a stringent protocol for viral sequence identification, such as VirusSeeker, that removes low-quality sequences, repeat sequences, and non-specific viral ‘hits.’ Similarly stringent methods have identified novel viral sequences (Zhao et al , 2013).…”