Performance and Agreement Between WGS Variant Calling Pipelines Used for Bovine Tuberculosis Control: Toward International Standardization

Abstract

Whole genome sequencing (WGS) and allied variant calling pipelines are a valuable tool for the control and eradication of infectious diseases, since they allow the assessment of the genetic relatedness of strains of animal pathogens. In the context of the control of tuberculosis (TB) in livestock, mainly caused by Mycobacterium bovis, these tools offer a high-resolution alternative to traditional molecular methods in the study of herd breakdown events. However, despite the increased use and efforts in the standardization of WGS methods in human tuberculosis around the world, the application of these WGS-enabled approaches to control TB in livestock is still in early development. Our study pursued an initial evaluation of the performance and agreement of four publicly available pipelines for the analysis of M. bovis WGS data (vSNP, SNiPgenie, BovTB, and MTBseq) on a set of simulated Illumina reads generated from a real-world setting with high TB prevalence in cattle and wildlife in the Republic of Ireland. The overall performance of the evaluated pipelines was high, with recall and precision rates above 99% once repeat-rich and problematic regions were removed from the analyses. In addition, when the same filters were applied, distances between inferred phylogenetic trees were similar and pairwise comparison revealed that most of the differences were due to the positioning of polytomies. Hence, under the studied conditions, all pipelines offer similar performance for variant calling to underpin real-world studies of M. bovis transmission dynamics.

How can DAFM best make use of whole genome sequencing to improve the effectiveness of the TB eradication programme?

“…For epidemiological studies, the SNP approach affords maximum genetic resolution to define clusters, and provides the basis for phylodynamic studies of directionality of infection and transmission rates both within-and between-species. Bespoke computational pipelines for the identification of SNPs in M. bovis populations have recently been compared, showing that under the tested conditions all pipelines offered similar performance for variant calling [11]; this provides confidence for comparison across studies. The main alternative to SNP-calling focuses on identifying variation at predefined loci (usually thousands of loci) and is an adaptation of multilocus sequencing typing (MLST).…”

“…Monaghan, and to use this as a route to development of computational approaches to integrate WGS with wildlife and cattle location data. The recently described tool for M. bovis variant calling and phylogenetic analysis is a recent output from the work [11].…”

Features of Mycobacterium bovis Complete Genomes Belonging to 5 Different Lineages

“…Indel and SNPs with ambiguous nucleotide present on at least one strain have also been filtered. SNPs present in PE/PPE family protein and pks12 were also filtered because of the low confidence and the higher error rate of these regions [ 48 , 49 , 50 ].…”