Method comparison of targeted influenza A virus typing and whole-genome sequencing from respiratory specimens of companion animals

Abstract

Epidemics of H3N8 and H3N2 influenza A viruses (IAVs) in dogs, along with recognition of spillover infections from IAV strains typically found in humans or other animals, have emphasized the importance of efficient laboratory testing. Given the lack of active IAV surveillance or immunization requirements for dogs, cats, or horses imported into the United States, serotype prediction and whole-genome sequencing of positive specimens detected at veterinary diagnostic laboratories are also needed. The conserved sequences at the ends of the viral genome segments facilitate universal amplification of all segments of viral genomes directly from respiratory specimens. Although several methods for genomic analysis have been reported, no optimization focusing on companion animal strains has been described, to our knowledge. We compared 2 sets of published universal amplification primers using 26 IAV-positive specimens from dogs, horses, and a cat. Libraries prepared from the resulting amplicons were sequenced using Illumina chemistry, and reference-based assemblies were generated from the data produced by both methods. Although both methods produced high-quality data, coverage profiles and base calling differed between the 2 methods. The sequence data were also used to identify the subtype of the IAV strains sequenced and then compared to standard PCR assays for neuraminidase types N2 and N8.

Keywords: M-RT-PCR; canine; equine; feline; influenza A virus; influenza matrix; multi-segment PCR; neuraminidase; reverse-transcription PCR; single-nucleotide polymorphism; typing; whole-genome sequencing.

Figure 1.

Gel image of M-RT-PCR product with the samples run with A. method A and B. method B conditions. Samples highlighted in red were chosen for sequencing.

Figure 2.

Fragment analyzer results of libraries prepared for sequencing using method A (A, B) and method B (C, D) conditions.

Figure 3.

Comparison of A. read quality scores, B. read count, and C. read length between methods. Values for method A are shown in gray and values for method B in white.

Figure 4.

Boxplot of normalized coverage for each segment of influenza A virus using both M-RT-PCR methods for A. H3N2 samples from set 1 and B. H3N8 samples from set 2. The dotted line is at 1, representing the mean coverage across the whole genome. Values for method A are shown in gray and method B in white.

Figure 5.

Relative coverage across each segment for set 1 H3N2 samples, with data for method A shown in blue and method B in orange. Each sample is represented by a light line, and the mean value for the method is shown as a dark line. The relative coverage for each base position is the number of reads at the base divided by the average coverage depth of the segment for a given sample.