Optimized Metavirome Analysis of Marine DNA Virus Communities for Taxonomic Profiling

Abstract

Recent advances in metavirome technology have provided new insights into viral diversity and function. The bioinformatic process of metavirome study is generally divided into two (or three) steps: assembly and taxonomic profiling including nucleotide alignment. Moreover, k-mer size and contig length are known to considerably affect the results of the assembly and consequently those of taxonomic profiling; however, the optimal k-mer size and contig length have not been established. In the present study, we analyzed marine virus DNA datasets with three different k-mer sizes using different assemblers: 1 k-mer (20) in the CLC Genomics Workbench, and 4 (21, 33, 55, and 77) and 5 (21, 33, 55, 77, and 99) k-mers in metaSPAdes. The use of large k-mers had the advantage of resolving more repeat regions, with higher N50 values and average contig lengths. The contig length helps reduce the error of continuous sequences and determine the number of viral operational taxonomic units. Our analysis suggested that 300 bp may be an appropriate minimum contig length, depending on the characteristics of viral samples. Based on the assembly result using metaSPAdes, we analyzed the DNA virus community using three taxonomic profiling tools: MG-RAST online server, the taxonomic profiling tools function in the CLC microbial module, and customized taxonomic assignment coding (CUTAXAC) using RStudio based on the BLASTn analysis. CUTAXAC showed the most diverse viral composition at the family and species levels along with the highest Shannon diversity index and fastest analysis time.