Centres for Doctoral Training

Probabilistic Genomics in the Era of Long Read Sequencing and Artificial Intelligence

Supervisors:

Arne Jacobs, MVLS, School of Biodiversty, One Health & Veterinary Medicine

Dirk Husmeier, College of Science & Engineering; School of Mathematics & Statistics

Maxwell Farrell, MVLS; School of Biodiversity, One Health & Veterinary Medicine

PhD Project Summary:

Modern genomics is generating vast amounts of sequencing data, but most analytical pipelines treat genetic variants as if they were known with certainty. In reality, sequencing errors, low coverage, and complex genomes introduce uncertainty that is often ignored, leading to biased inferences and reduced reliability. This challenge is particularly acute for long-read sequencing, ancient DNA, endangered species, and large-scale population studies.

This PhD will develop the next generation of uncertainty-aware genomics pipelines. The student will create new statistical and machine-learning methods that explicitly account for uncertainty throughout the analytical process, from variant discovery to downstream biological interpretation. Projects may include extending genotype-likelihood approaches to long-read sequencing, developing software for uncertainty-aware structural variant analysis, and designing deep-learning or large-language-model frameworks that incorporate probabilistic genetic information rather than fixed genotype calls.

The methods will be implemented as open-source software and evaluated using both simulated data and real genomic datasets from diverse species. The project sits at the interface of statistics, artificial intelligence, and genomics, offering comprehensive training in probabilistic modelling, machine learning, bioinformatics, and software development.

Graduates will gain highly transferable computational skills and contribute to a new generation of genomic analysis tools that deliver more reliable biological insights from imperfect data.