Sensitivity and specificity of metatranscriptomics as an arbovirus surveillance tool

Abstract

The ability to identify all the viruses within a sample makes metatranscriptomic sequencing an attractive tool to screen mosquitoes for arboviruses. Practical application of this technique, however, requires a clear understanding of its analytical sensitivity and specificity. To assess this, five dilutions (1:1, 1:20, 1:400, 1:8,000 and 1:160,000) of Ross River virus (RRV) and Umatilla virus (UMAV) isolates were spiked into subsamples of a pool of 100 Culex australicus mosquitoes. The 1:1 dilution represented the viral load of one RRV-infected mosquito in a pool of 100 mosquitoes. The subsamples underwent nucleic acid extraction, mosquito-specific ribosomal RNA depletion, and Illumina HiSeq sequencing. The viral load of the subsamples was also measured using reverse transcription droplet digital PCR (RT-ddPCR) and quantitative PCR (RT-qPCR). Metatranscriptomic sequencing detected both RRV and UMAV in the 1:1, 1:20 and 1:400 subsamples. A high specificity was achieved, with 100% of RRV and 99.6% of UMAV assembled contigs correctly identified. Metatranscriptomic sequencing was not as sensitive as RT-qPCR or RT-ddPCR; however, it recovered whole genome information and detected 19 other viruses, including four first detections for Australia. These findings will assist arbovirus surveillance programs in utilising metatranscriptomics in routine surveillance activities to enhance arbovirus detection.