HoloVir: A Workflow for Investigating the Diversity and Function of Viruses in Invertebrate Holobionts

Abstract

Abundant bioinformatics resources are available for the study of complex microbial metagenomes, however their utility in viral metagenomics is limited. HoloVir is a robust and flexible data analysis pipeline that provides an optimized and validated workflow for taxonomic and functional characterization of viral metagenomes derived from invertebrate holobionts. Simulated viral metagenomes comprising varying levels of viral diversity and abundance were used to determine the optimal assembly and gene prediction strategy, and multiple sequence assembly methods and gene prediction tools were tested in order to optimize our analysis workflow. HoloVir performs pairwise comparisons of single read and predicted gene datasets against the viral RefSeq database to assign taxonomy and additional comparison to phage-specific and cellular markers is undertaken to support the taxonomic assignments and identify potential cellular contamination. Broad functional classification of the predicted genes is provided by assignment of COG microbial functional category classifications using EggNOG and higher resolution functional analysis is achieved by searching for enrichment of specific Swiss-Prot keywords within the viral metagenome. Application of HoloVir to viral metagenomes from the coral Pocillopora damicornis and the sponge Rhopaloeides odorabile demonstrated that HoloVir provides a valuable tool to characterize holobiont viral communities across species, environments, or experiments.

Virome Assembly and Annotation: A Surprise in the Namib Desert

“…This degree of contamination was not evident when screening the reads against the complete SILVA database: 0.06% of the reads from the raw dataset and 0.1% of the reads from the filtered dataset mapped to rRNA genes at min bit-score 80, indicating negligible amounts of cellular contamination as defined by Roux et al (2013). Furthermore, at max 1.0E-10, only 577 of the 33296 predicted proteins (1.6%) mapped to cellular protein sequences from the extensive marker gene database published by Laffy et al (2016). …”

“…Previous studies with simulated datasets had not addressed these particular data issues. To identify a suitable assembly program, we simulated datasets that mimicked the copper-site data challenges, and compared performance of five assemblers that had previously shown best results in metagenome reconstruction (de Cárcer et al, 2014; Vázquez-Castellanos et al, 2014; García-López et al, 2015; Laffy et al, 2016). Of particular interest was the assembly precision for different size sequences, which are expected in metagenomes.…”

“…Both, the 13.9 M raw reads, as well as the 6.9 M quality filtered reads were mapped to the SILVA database using BLASTn (BLAST+) using a min bit-score threshold of 80. The extensive dataset of viral and cellular marker proteins (Laffy et al, 2016), was downloaded from XXX, and all proteins predicted on the final copper site assembly were mapped using BLASTp (BLAST+) with a maximum e -value threshold of 1.0e-10.…”

“…Several metagenome assemblers that account for varying sequencing depths in the data have been developed (e.g., Peng et al, 2011; Bankevich et al, 2012; Boisvert et al, 2012). Their suitability for virome assembly has been assessed in recent studies (de Cárcer et al, 2014; Vázquez-Castellanos et al, 2014; García-López et al, 2015; Laffy et al, 2016). This was effectively done using simulated data: virtual viromes consisting of known viral genomes at varying abundance levels serve to generate synthetic reads, which are reassembled into contigs and scaffolds.…”

“…Diverse methods (e.g., filtering for 16S rRNA and tRNA) and tools (e.g., Schmieder and Edwards, 2011; Roux et al, 2015) for prescreening virome data have been applied. The recently published virome annotation program HoloVir (Laffy et al, 2016), addresses this problem by prescreening the sequences against the entire SILVA database and an extensive database with viral and cellular marker genes. However, comparative analyses assessing the efficiency of these procedures are still outstanding.…”