Baillie lab, Roslin Institute, University of Edinburgh

Translational genomics in critical care medicine

We are a cross-disciplinary translational research group focused on using genomics and transcriptomics to better understand and treat critical illness. We develop and apply computational tools, and use in vitro and in vivo models to generate and test hypotheses. We are based in the Roslin Institute, University of Edinburgh and Intensive Care Unit, Royal Infirmary Edinburgh.

Read a summary of our approach in this perspective article:

Baillie, J. K. Science 344, 807–808 (May 23, 2014).

The GenOMICC Study: recruiting now

Infectious diseases and severe injuries affect millions of people around the world every year. Most cases are mild, but some people develop life-threatening illness. By finding the genes that determine severe illness, we aim to develop new treatments that can help people to survive.

The GenOMICC study ( Genetics Of Mortality and susceptibility In Critical Care ) is looking for volunteers to give us a DNA sample to help us to understand why some people become critically ill with infections and injuries. By finding the genes that determine severe illness, we aim to develop new treatments that can help people to survive.

Applications of computational biology in critical care medicine.

Stratified medicine

Throughout the history of medicine, progress has been made by recognizing patterns of disease, or syndromes. When new technologies are invented, they can reveal observable characteristics that have close relationships to disease trajectories, outcomes, and most importantly of all, response to therapy. Genomics and transcriptomics have these properties, but raise an additional challenge: the scale of the data is such that detecting relevant signals is a substantial computational and mathematical challenge.

Russell, C. D. & Baillie, J. K. Current Opinion in Systems Biology 2, 139–145 (Apr. 2017).
Gene set hypothesis testing.

Gene set hypothesis testing

The lack of specific hypotheses in traditional genome-wide association studies has a disadvantage - it greatly limits the statistical power, ensuring that only the strongest signals can be detected, even with very large numbers of subjects.

Gene set hypothesis testing (GSHT) enables investigators to take a biological hypothesis and test it using GWAS data. For example, in our paper, we tested the hypothesis that failure to downregulate inflammatory signals during macrophage maturation is a key mechanism in the pathogenesis of Crohn's disease. We defined search criteria to find inflammatory genes that are downregulated during macrophage differentiation, and then empirically tested the hypothesis that these genes are likely to be associated with Crohn's disease.

Baillie, J. K. et al. PLOS Genetics 13, e1006641 (Mar. 6, 2017).


In the FANTOM5 consortium we showed that, with a sufficiently rich expression dataset, it is possible to infer functional relationships between genes (and other transcribed elements in the genome, such as enhancers and lncRNAs) by grouping them according to similarities in expression. We proved this using known pathways, but the real value of this finding is that it enables us to discover relationships between genomic regions about which almost nothing is already known.

Forrest*, Kawaji*, Rehli*, Baillie*, et al. Nature 507, 462–470 (Mar. 27, 2014).

FANTOM5 coexpression network


We are very grateful to recieve funding from the following sources: Wellcome Trust, BBSRC, MRC, NIH.

Contact us