Viromes vs. mixed community metagenomes: choice of method dictates interpretation of viral community ecology

Abstract

Viruses, the majority of which are uncultivated, are among the most abundant biological entities in microbiomes and ecosystems on Earth. From selfishly altering community-wide microbial physiology to driving microbiome dynamics, viruses (particularly bacteriophage) are fundamental members of microbiomes. While the number of studies leveraging viral metagenomics (viromics) for studying uncultivated viruses is growing, standards for viromics research are lacking. Viromics can leverage computational discovery of viruses from total metagenomes of all community members (hereafter metagenomes) or use physical separation of virus-specific fractions (hereafter viromes). However, differences in the recovery and interpretation of viruses from metagenomes and viromes obtained from the same samples remains understudied. Here, we compare viral communities from paired viromes and metagenomes obtained from 51 diverse samples across human gut, soil, freshwater, and marine ecosystems. Overall, viral communities obtained from viromes were more abundant and species rich than those obtained from metagenomes, although there were some exceptions. Despite this, metagenomes still contained many viral genomes not detected in viromes. We also found notable differences in the predicted lytic state of viruses detected in viromes vs metagenomes at the time of sequencing. Other forms of variation observed include genome presence/absence, genome quality, and encoded protein content between viromes and metagenomes, but the magnitude of these differences varied by environment. Overall, our results here show that choice of method can lead to differing interpretations of viral community ecology. We suggest that the choice of whether to target a metagenome or virome should be dependent on the environmental context and ecological questions being asked. However, our overall recommendation to researchers investigating viral ecology and evolution is to pair both approaches to maximize their respective benefits.

Targeted Viromes and Total Metagenomes Capture Distinct Components of Bee Gut Phage Communities

“…Here we compare how two common forms of phage community sampling, bioinformatic mining of total metagenomes and targeted sequencing of viromes, influence the phage communities recovered from the guts of two commercially important bee species: Apis mellifera and Bombus impatiens. Similar to the results of previous studies [16][17][18] , we show that sampling method significantly affected alpha diversity, beta diversity, composition, and structure of sampled phage communities. In particular, we show that viromes outperform total metagenomes in terms of the number of phage recovered, but that this can depend on the specific environment being sampled.…”

“…While our results highlight the ability of total metagenomes to recover phage missed by viromes, they also showcase the capacity of viromes to sample a greater diversity of phage. Similar to previous comparisons of viromes and total metagenomes in human gut, soil, and freshwater environments 17,18,29 , our honey bee viromes recovered a substantially larger number of phage than did total metagenomes. Further, the occupancy-abundance relationships examined here and by Medellin et al 16 show that total metagenomes tend to be biased towards sampling the most abundant and prevalent phage, while viromes are more successful at sampling rare phage.…”

“…With more sequencing devoted to viruses, targeted viromes can recover a greater diversity of phage relative to total metagenomes. This is illustrated by previous comparisons of viromes and total metagenomes in soil and freshwater samples 16,17 . However, whether viromes always outperform total metagenomes is not clear.…”

“…As such, insights gained from studying free-living phage communities may not always apply to communities of host-associated phage. Specifically, while the best methods for sampling free-living phage communities have been empirically compared 16,17 , little work has evaluated the best methods for sampling host-associated phage 18 . Moreover, given the importance of microbiomes to host health, understanding phage community dynamics and how phage differ between related animal hosts is a key priority.…”

“…However, whether viromes always outperform total metagenomes is not clear. Work applying these two methods to host-associated (human gut) and low-biomass (deep-sea marine) samples has produced conflicting results 17,18 . One explanation for these discrepancies is the biases associated with total metagenomes and targeted viromes 30,31 .…”