Portable Rabies Virus Sequencing in Canine Rabies Endemic Countries Using the Oxford Nanopore MinION

Abstract

As countries with endemic canine rabies progress towards elimination by 2030, it will become necessary to employ techniques to help plan, monitor, and confirm canine rabies elimination. Sequencing can provide critical information to inform control and vaccination strategies by identifying genetically distinct virus variants that may have different host reservoir species or geographic distributions. However, many rabies testing laboratories lack the resources or expertise for sequencing, especially in remote or rural areas where human rabies deaths are highest. We developed a low-cost, high throughput rabies virus sequencing method using the Oxford Nanopore MinION portable sequencer. A total of 259 sequences were generated from diverse rabies virus isolates in public health laboratories lacking rabies virus sequencing capacity in Guatemala, India, Kenya, and Vietnam. Phylogenetic analysis provided valuable insight into rabies virus diversity and distribution in these countries and identified a new rabies virus lineage in Kenya, the first published canine rabies virus sequence from Guatemala, evidence of rabies spread across an international border in Vietnam, and importation of a rabid dog into a state working to become rabies-free in India. Taken together, our evaluation highlights the MinION's potential for low-cost, high volume sequencing of pathogens in locations with limited resources.

Keywords: MinION; canine rabies elimination; lyssavirus; nanopore; portable sequencing; rabies.

Figure 1

Errors observed in consensus sequences generated from MinION sequencing data by three methods. Number of differences in the nucleoprotein gene (1353 nt) for raw consensus (Raw Consensus), consensus after polishing with raw reads using Nanopolish (Polished Consensus), and after manual correction of indels in homopolymer regions (Manual Correction) compared to consensus generated by Sanger sequencing is shown. Insertions, deletions, and single nucleotide changes (miscalled bases or SNPs) are shown in different shades of blue. Number of each type of error is depicted on the bars. Data shown include data from all sequences (left) and those sequences produced with >50x read depth (right).

Figure 2

Phylogenetic analysis of MinION sequences generated in Goa, India. Complete nucleoprotein gene sequences from Goa rabies virus isolates were compared to publicly available sequences. Reference sequences were chosen based on sequence similarity to the newly generated sequences or inclusion in rabies Arctic-like 1, Arctic, Arctic-like 2, Arctic-like 3, Indian Subcontinent, and Cosmopolitan phylogenetic clades (from [31]). Newly generated sequences are shown in bold; several sequences are collapsed for viewing ease. Sequences from Goa isolates are shown in magenta. Translocation case from Rajasthan is highlighted in green. Accession number, host, location, and collection date are included for reference, when available; additional sample details can be found in Table S1. Phylogenetic analysis was performed by maximum likelihood based on a GTR+G+I model. Differences between samples are shown by the number of changes per site along the horizontal axis. Bootstrap values near the branch points represent the percentage of trees that had the same clustering out of 1000 replicates.

Figure 3

Phylogenetic analysis of MinION sequences generated in Nairobi, Kenya. Full-length glycoprotein gene sequences (1575 nt) from Kenya rabies virus isolates were compared to publicly available sequences. Reference sequences were chosen based on sequence similarity to the new Kenya sequences or inclusion in Africa 1a, Africa 1b, Africa 1c, Africa-2, Africa-3, or Africa 4 rabies phylogenetic clades based on Troupin et al. [31]. Accession number, location, and collection date are included, when available. Newly generated sequences are highlighted in color, corresponding to rough location on the map to the left. Additional sample information can be found in Table S1. Phylogenetic analysis was performed by maximum likelihood based on the GTR+G+I model. Number of changes per site is shown along the horizontal axis. Bootstrap values near the branch points represent the percentage of trees that had the same clustering out of 1000 replicates.

Figure 4

Phylogenetic analysis of MinION sequences generated in Guatemala City, Guatemala. Complete nucleoprotein gene sequences from Guatemala rabies virus isolates were compared to publicly available sequences. Reference sequences were chosen based on sequence similarity to the new Guatemala sequences, isolation from vampire bat or bovine in Mexico or Guatemala, or inclusion Cosmopolitan Americas-2a (AM2a) rabies phylogenetic clade based on Troupin et al. [31]. Accession number, host animal, location, and collection year are included for reference, when available. Sequences generated in this study are highlighted in bold; additional information can be found in Table S1. Colored points on the map correspond to the location of LN34 positive samples; confirmed canine or vampire bat lineage isolates are colored dark blue or dark pink on the map and phylogenetic tree. Suspected lineage was based on host (canine or bovine). Phylogenetic analysis was performed by maximum likelihood based on the GTR+G+I model in Mega7. Number of changes per site is shown along the horizontal axis. Bootstrap values near the branch points represent the percentage of trees that had the same clustering out of 1000 replicates.

Figure 5

Phylogenetic analysis of MinION sequences generated in Hanoi, Vietnam. Complete nucleoprotein gene sequences from Vietnam rabies virus isolates were compared to publicly available sequences. Reference sequences were chosen based on sequence similarity to the new Vietnam sequences or inclusion in South East Asia 1–5 (SEA1–SEA5) or Cosmopolitan rabies phylogenetic clades based on Troupin et al. [31]. Sequences generated in this study are highlighted in bold; some sequences have been collapsed for ease of viewing. Color coding of sequence names correspond to region location in Vietnam as colored on the map. Accession number, host animal, location, and collection date are included for reference, when available; additional information can be found in Table S1. Differences between samples are shown by the number of changes per site along the horizontal axis. Bootstrap values near the branch points represent the percentage of trees that had the same clustering out of 1000 replicates.