Using machine learning improves predictions of herd-level bovine tuberculosis breakdowns in Great Britain.

In the United Kingdom, despite decades of control efforts, bovine tuberculosis (bTB) has not been controlled and currently costs ~ £100 m annually. Critical in the failure of control efforts has been the lack of a sufficiently sensitive diagnostic test. Here we use machine learning (ML) to predict herd-level bTB breakdowns in Great Britain (GB) with the aim of improving herd-level diagnostic sensitivity. The results of routinely-collected herd-level tests were correlated with risk factor data. Four ML methods were independently trained with data from 2012-2014 including ~ 4700 positive herd-level test results annually. The best model's performance was compared to the observed sensitivity and specificity of the herd-level test calculated on the 2015 data resulting in an increased herd-level sensitivity from 61.3 to 67.6% (95% confidence interval (CI): 66.4-68.8%) and herd-level specificity from 90.5 to 92.3% (95% CI: 91.6-93.1%). This approach can improve predictive capability for herd-level bTB and support disease control.

The evolution and phylodynamics of serotype A and SAT2 foot-and-mouth disease viruses in endemic regions of Africa.

Foot-and-mouth disease (FMD) is a major livestock disease with direct clinical impacts as well as indirect trade implications. Control through vaccination and stamping-out has successfully reduced or eradicated the disease from Europe and large parts of South America. However, sub-Saharan Africa remains endemically affected with 5/7 serotypes currently known to be circulating across the continent. This has significant implications both locally for livestock production and poverty reduction but also globally as it represents a major reservoir of viruses, which could spark new epidemics in disease free countries or vaccination zones. This paper describes the phylodynamics of serotypes A and SAT2 in Africa including recent isolates from Cameroon in Central Africa. We estimated the most recent common ancestor for serotype A was an East African virus from the 1930s (median 1937; HPD 1922-1950) compared to SAT2 which has a much older common ancestor from the early 1700s (median 1709; HPD 1502-1814). Detailed analysis of the different clades shows clearly that different clades are evolving and diffusing across the landscape at different rates with both serotypes having a particularly recent clade that is evolving and spreading more rapidly than other clades within their serotype. However, the lack of detailed sequence data available for Africa seriously limits our understanding of FMD epidemiology across the continent. A comprehensive view of the evolutionary history and dynamics of FMD viruses is essential to understand many basic epidemiological aspects of FMD in Africa such as the scale of persistence and the role of wildlife and thus the opportunities and scale at which vaccination and other controls could be applied. Finally we ask endemic countries to join the OIE/FAO supported regional networks and take advantage of new cheap technologies being rolled out to collect isolates and submit them to the World Reference Laboratory.

Analysis of bovine viral diarrhoea virus: Biobank and sequence database to support eradication in Scotland.

Samples from bovine viral diarrhoea virus (BVDV)-positive cattle were gathered by Scottish diagnostic laboratories and used to produce a Biobank of samples with associated location and identification data in support of the Scottish BVDV eradication scheme. The samples were subject to direct amplification and sequencing of the 5'-untranslated region (5'-UTR) to define the viral types and subtypes present. From 2693 samples collected prior to 2016, approximately 2300 sequences were obtained, representing 8 BVDV type 1 subtypes. No BVDV type 2 samples were detected. The samples came from all regions of the UK but 66 per cent were from Scotland. Analysis of the sequences showed great diversity in the 5'-UTR, with 1206 different sequences. Many samples carried virus with identical 5'-UTR sequences; often from single locations, but there were also examples of the same sequence being obtained from samples at several different locations. This work provides a resource that can be used to analyse the movement of BVDV strains both within Scotland and between Scotland and other nations, particularly in the latter stages of the Scottish eradication programme, and so inform the advice available to both livestock keepers and policymakers.

Bones hold the key to DNA virus history and epidemiology.

DNA in human skeletal remains represents an important historical source of host genomic information and potentially of infecting viruses. However, little is known about viral persistence in bone. We searched ca. 70-year-old long bones of putative Finnish casualties from World War II for parvovirus B19 (B19V) DNA, and found a remarkable prevalence of 45%. The viral sequences were exclusively of genotypes 2 (n = 41), which disappeared from circulation in 1970´s, or genotype 3 (n = 2), which has never been reported in Northern Europe. Based on mitochondrial and Y-chromosome profiling, the two individuals carrying B19V genotype 3 were likely from the Soviet Red Army. The most recent common ancestor for all genotypes was estimated at early 1800s. This work demonstrates the forms of B19V that circulated in the first half of the 20(th) century and provides the first evidence of the suitability of bone for exploration of DNA viruses.

Real-time characterization of the molecular epidemiology of an influenza pandemic.

Early characterization of the epidemiology and evolution of a pandemic is essential for determining the most appropriate interventions. During the 2009 H1N1 influenza A pandemic, public databases facilitated widespread sharing of genetic sequence data from the outset. We use Bayesian phylogenetics to simulate real-time estimates of the evolutionary rate, date of emergence and intrinsic growth rate (r0) of the pandemic from whole-genome sequences. We investigate the effects of temporal range of sampling and dataset size on the precision and accuracy of parameter estimation. Parameters can be accurately estimated as early as two months after the first reported case, from 100 genomes and the choice of growth model is important for accurate estimation of r0. This demonstrates the utility of simple coalescent models to rapidly inform intervention strategies during a pandemic.

The evolutionary dynamics of influenza A virus adaptation to mammalian hosts.

Few questions on infectious disease are more important than understanding how and why avian influenza A viruses successfully emerge in mammalian populations, yet little is known about the rate and nature of the virus' genetic adaptation in new hosts. Here, we measure, for the first time, the genomic rate of adaptive evolution of swine influenza viruses (SwIV) that originated in birds. By using a curated dataset of more than 24 000 human and swine influenza gene sequences, including 41 newly characterized genomes, we reconstructed the adaptive dynamics of three major SwIV lineages (Eurasian, EA; classical swine, CS; triple reassortant, TR). We found that, following the transfer of the EA lineage from birds to swine in the late 1970s, EA virus genes have undergone substantially faster adaptive evolution than those of the CS lineage, which had circulated among swine for decades. Further, the adaptation rates of the EA lineage antigenic haemagglutinin and neuraminidase genes were unexpectedly high and similar to those observed in human influenza A. We show that the successful establishment of avian influenza viruses in swine is associated with raised adaptive evolution across the entire genome for many years after zoonosis, reflecting the contribution of multiple mutations to the coordinated optimization of viral fitness in a new environment. This dynamics is replicated independently in the polymerase genes of the TR lineage, which established in swine following separate transmission from non-swine hosts.

Estimating reassortment rates in co-circulating Eurasian swine influenza viruses.

Swine have often been considered as a mixing vessel for different influenza strains. In order to assess their role in more detail, we undertook a retrospective sequencing study to detect and characterize the reassortants present in European swine and to estimate the rate of reassortment between H1N1, H1N2 and H3N2 subtypes with Eurasian (avian-like) internal protein-coding segments. We analysed 69 newly obtained whole genome sequences of subtypes H1N1-H3N2 from swine influenza viruses sampled between 1982 and 2008, using Illumina and 454 platforms. Analyses of these genomes, together with previously published genomes, revealed a large monophyletic clade of Eurasian swine-lineage polymerase segments containing H1N1, H1N2 and H3N2 subtypes. We subsequently examined reassortments between the haemagglutinin and neuraminidase segments and estimated the reassortment rates between lineages using a recently developed evolutionary analysis method. High rates of reassortment between H1N2 and H1N1 Eurasian swine lineages were detected in European strains, with an average of one reassortment every 2-3 years. This rapid reassortment results from co-circulating lineages in swine, and in consequence we should expect further reassortments between currently circulating swine strains and the recent swine-origin H1N1v pandemic strain.

Detection of mammalian virulence determinants in highly pathogenic avian influenza H5N1 viruses: multivariate analysis of published data.

Highly pathogenic avian influenza (HPAI) virus H5N1 infects water and land fowl and can infect and cause mortality in mammals, including humans. However, HPAI H5N1 strains are not equally virulent in mammals, and some strains have been shown to cause only mild symptoms in experimental infections. Since most experimental studies of the basis of virulence in mammals have been small in scale, we undertook a meta-analysis of available experimental studies and used Bayesian graphical models (BGM) to increase the power of inference. We applied text-mining techniques to identify 27 individual studies that experimentally determined pathogenicity in HPAI H5N1 strains comprising 69 complete genome sequences. Amino acid sequence data in all 11 genes were coded as binary data for the presence or absence of mutations related to virulence in mammals or nonconsensus residues. Sites previously implicated as virulence determinants were examined for association with virulence in mammals in this data set, and the sites with the most significant association were selected for further BGM analysis. The analyses show that virulence in mammals is a complex genetic trait directly influenced by mutations in polymerase basic 1 (PB1) and PB2, nonstructural 1 (NS1), and hemagglutinin (HA) genes. Several intra- and intersegment correlations were also found, and we postulate that there may be two separate virulence mechanisms involving particular combinations of polymerase and NS1 mutations or of NS1 and HA mutations.