Bioinformatics Seminars

Bioinformatics Seminar

Time: 10:45am Tuesdays.
Level 7 Seminar Room 2, WEHI1

3 July 2018

Improved prediction of celiac disease using known HLA risk haplotypes for improved biological and clinical insight

Benjamin Goudey
IBM Research

Celiac disease (CD) is an autoimmune condition that affects genetically susceptible people whereby the small intestine becomes inflamed when gluten is consumed. Genetic testing of CD based on genes in the Human Leukocyte Antigen (HLA) is useful in the clinical work-up of CD as a negative result almost guarantees you don't have CD. However, the test's low positive predictive value has led research efforts to develop genetic risk scores (GRS) that make use of genetic variation outside of the HLA region to improve the accuracy of predictive modelling. We hypothesised that the predictive power of HLA haplotype information that is currently clinically collected may be greater than previously reported.

In this talk, I'll describe a novel GRS for CD that only relies on HLA information that is currently collected clinically but still achieves state-of-the-art predictive performance. Conditioning on our proposed model, we find two novel non additive associations (HLA-DQ6.2, HLA-DQ7.3 with the known CD risk allele HLA-DQ2.5). Finally, I will discuss how this model may impact clinical CD testing, the relevance of our insights for other immune-related conditions and for interpretation of GRS.

Search past seminars