Aquatic viruses heavily impact their microbial hosts, as they lyse up to 40% of all ocean microbes each day, contribute novel genetic material, and alter host metabolisms. Viruses also influence ecosystem functioning, as aquatic microbes are the base of the food web and drive biogeochemical cycling, including carbon sequestration. Little is known about the presence and diversity of viruses, due to the low efficiency in culturing and the lack of a universal marker gene. Although next generation sequencing of metagenomes has greatly increased our depth of knowledge on cultured and unculturable viruses in the environment, non-viral (cellular) genetic material contaminates all viral nucleic acid extractions. Several studies reduce cellular DNA contamination by chloroform “washing” samples in combination with nuclease treatments. Others rely on in silico approaches to big data to separate viral from cellular sequences once the data are acquired. Yet, the consequences of these approaches on viral community representation have not been systematically evaluated. We have designed a study to assess the efficacy of in vitro decontamination (DNaseI, chloroform) with in silico sequence classification methods (combinations of CheckM, VirSorter/VirFinder, centrifuge) using aquatic virus samples. Five ocean virus samples were processed using established FeCl3 flocculation and resuspension. Prior to DNA extraction, the resuspensions were equally split into experimental treatments to reduce cellular DNA: 1) chloroform, 2) DNase I, 3) chloroform followed by DNase I, or 4) no treatment.
Results/Conclusions
A total of 286,008,356 151-base pair reads were analyzed. Treatment with chloroform and DNaseI resulted in the highest proportion of reads classified as viral by a custom centrifuge index (0.44±0.14), followed by chloroform (0.34±0.14), no treatment (0.26±0.13), and DNaseI only (0.24±0.10) treatments. On average the number of virus-classified bins: (1) increased by 12.6% with chloroform treatment alone, (2) decreased by 10.2% with DNase I treatment alone, and (3) decreased by 1.6% with the combined chloroform and DNase I treatment, in comparison to no treatment. Total viral genome sizes in base pairs decreased by 11.4%, 16.6%, and 25.7% over the same three treatments. Mean assembly length in viral bins was reduced 71.7%, 15.5%, and 59.5%, respectively. These preliminary results indicate that in vitro treatments affect not only the quantity of viruses identified from an environmental sample via metagenomics, but also the quantity of the metagenomic data available for each virus.