68 R.A. Skuce et al.
we consider the opportunities and limitations
presented by the game-changing revolution in
pathogen genomic epidemiology (Kwong et al.,
2015).
Technology does not stand still for long and
the revolution in whole-genome sequencing
(WGS) is transforming pathogen molecular epi-
demiology and will ultimately provide portable,
high-performance tests for outbreak investiga-
tion and epidemiological research. Increasingly
sophisticated analytical tools that combine the
whole-genome sequences of large samples of
pathogen isolates with geospatial and temporal
data are leading to unprecedented improve-
ments in understanding of the origins, evolution,
pathogenicity and epidemiology of infectious
diseases, with the prospect of greatly improved
real-time surveillance of disease outbreaks
(Bentley and Parkhill, 2015; Croucher and
Didelot, 2015). Following rapid developments in
bacterial WGS, it is now cost effective to read and
compare the complete genome of bacteria; this
provides the highest possible discrimination
between isolates and enhanced resolution of
complex epi-systems (Luheshi et al., 2015; Lee
and Behr, 2016). Mutations which accrue as
bacteria spread between hosts allow scientists to
produce and compare detailed pathogen family
trees and to compare these to the geographical
distribution of the bacteria.
Several recent studies on human TB
genomic epidemiology illustrate particularly
well the power and, indeed, some of the limita-
tions of this phylodynamic approach to investi-
gating TB transmission dynamics (Gardy et al.,
2011; Walker et al., 2013a, 2013b, 2014; Dide-
lot et al., 2014; Tang and Gardy, 2014; Guthrie
and Gardy, 2015; Nikolayevskyy et al., 2016). To
some extent, such studies are confounded by
missing data and the exceptionally slow muta-
tion rate characteristic of this pathogen (Ford
et al., 2013; Lillebaek et al., 2016). It has proven
important to estimate the mutation rate of the
M. tuberculosis lineage(s) under study, to more
fully understand within-host diversity (Thacker
et al., 2015; Didelot et al., 2016) and, uniquely,
to calibrate and date-stamp the transmission
trees derived (Ford et al., 2013; Colijn and Gardy,
2014; Didelot et al., 2014; Thacker et al., 2015)
when attempting to investigate transmission
dynamics using genomic epidemiology. Increas-
ingly advanced statistical and epidemiological
modelling approaches are required to more fully
exploit TB phylodynamics (Didelot, 2013; Coll
et al., 2014b; Didelot et al., 2014; Jombart et al.,
2014a, 2014b). However, bacterial WGS is
increasingly being seen as an enabling technol-
ogy for the high-resolution microbial forensic
analysis of pathogens and mycobacterial molec-
ular surveillance systems are being implemented
(Pankhurst et al., 2016).
A pilot study in Northern Ireland investi-
gated the potential of molecular typing, bacterial
WGS and mathematical modelling to investigate
how bovine TB might spread between and within
cattle herds and local wildlife (Biek et al., 2012).
Previous MLVA surveillance had shown that
cattle and badgers were associated at a regional
scale, but lacked genetic evidence linking cattle
and badgers at the farm scale. The pilot investi-
gated a local TB ‘micro-epidemic’ comprising
five cattle herds with a 10-year history of
repeated TB breakdowns (due to a novel molecu-
lar type, MLVA010) and four TB-positive road
traffic accident (RTA) badgers. Bacterial whole-
genomes were sequenced from a sample of the
TB-positive cattle (n = 26) and the four RTA bad-
gers. The study showed the following: (i) most
breakdowns involved genetically distinct bacte-
ria; and (ii) cattle and local badger bacteria were
very highly related and often indistinguishable,
implying that transmission between cattle and
badgers occurred frequently, recently and at a
very local (farm-level) scale. This study produced
the first direct genetic evidence of ongoing
bovine TB transmission between cattle and bad-
gers at the individual farm level. However, the
number of isolates examined was too low to
draw robust conclusions about the direction of
transmission between cattle and badgers at that
stage. Results were also consistent with some
ongoing cattle–cattle transmission (amplifica-
tion) occurring in some study herds. Recorded
cattle movements, between the five study herds,
did not seem to be significant in determining the
geographical distribution of the pathogen. How-
ever, since sampling was localized, movement
effects may have been underestimated.
Further isolates were recovered from
the expanding micro-epidemic of MLVA010
(Trewby et al., 2016). As before, this micro-
epidemic was locally driven and recorded cattle
movements were not a strong predictor of the
distribution of the bacteria, which do spread in