A light-hearted look at probability and statistics on Newstalk, joining Seán Moncrieff on the Moncrieff Show to unpack empirical coin-tossing experiments and perceived fairness.

• Listen back to the segment: How fair is a coin toss actually?

How group structure impacts the numbers at risk for coronary artery disease: polygenic risk scores and nongenetic risk factors in the UK Biobank cohort

We are delighted that this paper was selected by GENETICS as a July 2024 Issue Highlight / Feature Article.

Jinbo successfully defended her thesis, “Statistical Genetics Models with Actuarial Applications”, before external examiner Dr. Peter Joshi and internal examiner Prof. Thomas Brendan Murphy, passing with minor corrections.

Many thanks to Dr. Adrian O’Hagan for co-supervising Jinbo throughout her PhD, and to Dr. Isabella Gollini for chairing the examination.

Kimberly Conteddu, Holly M. English, Andrew W. Byrne, Bawan Amin, Laura L. Griffin, Prabhleen Kaur, Virginia Morera-Pujol, Kilian J. Murphy, Michael Salter-Townshend, Adam Francis Smith, and Simone Ciuti examine multi-host zoonotic epidemiology in A scoping review on bovine tuberculosis highlights the need for novel data streams and analytical approaches to curb zoonotic diseases.

• Background: Zoonotic diseases represent a major societal challenge in terms of health and economic impacts. One Health approaches to managing zoonoses require cross-disciplinary collaboration, novel data streams, and statistical modelling. Bovine tuberculosis (bTB) is a prime example involving domestic livestock, wildlife hosts, and environmental factors.
• Objective: We undertook a scoping review of multi-host bTB epidemiology to identify trends in host species focus, analytical methodologies, spatial scales, and One Health integration.
• Results: Reviewing 167 articles, we found strong geographic and species concentration on the badger–cattle episystem, with far less attention to multi-host systems in southern Africa. Relatively few studies modelled management exit strategies, human disturbance, habitat change, or weather/climate variables.
• Conclusions: Linking high-resolution empirical wildlife tracking with mathematical and simulation modelling offers a “virtuous cycle” to inform bTB policy and broader zoonotic disease control.

A simulation study for multifactorial genetic disorders to quantify the impact of polygenic risk scores on critical illness insurance

Polygenic risk scores (PRS) assess inherited predisposition to complex diseases and have two primary utilities relevant to insurance: identifying high-risk groups alongside standard clinical factors, and informing early preventive interventions. Using coronary heart disease as a case study under a liability threshold model, we introduce PRS into an actuarial modelling framework to quantify adverse selection arising from information asymmetry. Exploring three applicant behaviour scenarios, we show that PRS predictive accuracy and the proportion of applicants aware of their genetic risk have the largest impact on equilibrium insurance premiums.

Prabhleen Kaur, Simone Ciuti, Michael Salter-Townshend, and co-authors present A protocol for assessing bias and robustness of social network metrics using GPS based radio-telemetry data.

In wildlife studies, researchers can typically tag only a subset of individuals in a population, inferring broader social structure from that sample. Using empirical telemetry datasets across five ungulate species in diverse environments, we evaluate how robust global and local network summaries are to downsampling, autocorrelation, and population density—and provide both a step-by-step statistical protocol and open-source software (aniSNA) to carry out each stage of the analysis.

Genome-wide local ancestry and evidence for mitonuclear coadaptation in African hybrid cattle populations (Bos taurus/indicus)

Applying local ancestry inference across admixed African cattle breeds, we find significantly elevated taurine nuclear ancestry at genes whose protein products interact directly with mitochondrially encoded proteins—providing genome-wide functional evidence for mitonuclear coadaptation in hybrid mammals.

@article{2022,
  author = {},
  title = {Genome-Wide {Local} {Ancestry} and {Mitonuclear}
    {Coadaptation} in {African} {Hybrid} {Cattle}},
  journal = {iScience},
  date = {2022-06-27},
  url = {https://doi.org/10.1016/j.isci.2022.104672},
  doi = {10.1016/j.isci.2022.104672},
  langid = {en}
}

Static and dynamic methods in social network analysis reveal the association patterns of desert-dwelling giraffe

Computational modelling of chromosomally clustering protein domains in bacteria

Together with Dr. Niamh Cahill and Prof. Andrew Parnell, I co-convened a session on Bayesian statistical analysis methods for paleoclimate data. The session drew a full room with six invited and contributed talks alongside an active poster session, highlighting how hierarchical Bayesian methodology tackles inverse problems such as climate reconstruction under multiple sources of chronological and proxy uncertainty.

I was honoured to deliver a keynote lecture on Social Network Analysis to conservation biologists and movement ecologists from 33 research centres and environmental institutes at the Eurodeer XI meeting in Dublin. The network brings researchers together annually to harmonise telemetry datasets and collaborate on ecological studies.

Following the workshop, our collaboration applies statistical network models—including stochastic blockmodels, role analysis, and latent space approaches—to multi-site GPS telemetry across diverse terrain, climate, and vegetation regimes, quantifying how supplemental feeding stations affect movement patterns and social interactions in wild deer populations.

This was great fun to take part in! Futureproof is Newstalk’s science programme hosted by Jonathan McCrea. A special feature episode on Chance and Coincidence (“Chances”) was recorded with Prof. Brian Hughes (Professor of Psychology at the University of Galway) and me.

Brian explored why humans naturally seek patterns and assign meaning to random events, after which I worked through back-of-the-envelope probability calculations for coincidence stories sent in by listeners. Because the marginal probability of any hyper-specific sequence of events is tiny once exact times and incidental details are included, I stripped each story back to a baseline “non-coincidence” version and estimated the conditional odds of the surprising overlap that made the story worth telling. Hopefully the result was entertaining, illuminating, or perhaps both!

• Listen to the episode: Futureproof — Chances

• SFI CRT in Foundations of Data Science
• SFI CRT in Genomics Data Science

Both programmes offer fully funded cohort-based PhD training across data science, statistical methodology, and genomics. Follow the links above for programme and application details.

BEEF-SIG is a four-year collaborative research project between University College Dublin (UCD) and Teagasc (the Irish Agriculture and Food Development Authority), funded by the Department of Agriculture, Food and the Marine starting 1 January 2019.

Irish beef is internationally recognised for its pasture-based production—for example, export markets such as the United States seek verification of animals raised on a predominantly grass diet under Bord Bia’s Sustainable Beef & Lamb Assurance Scheme (SBLAS). BEEF-SIG combines stable isotope ratio analysis, trace elemental profiling, and multivariate statistical classification to objectively quantify farm feed inputs (grass, silage, concentrates) and verify grass-fed origin claims.