Immunogenetics and TCR expression patterns together predict future T1D onset from multi-omic datasets

Type 1 Diabetes (T1D) is an autoimmune disease affecting juvenile populations across the world. Progression to symptomatic disease, currently predicted by ancestral risk and the presence of islet reactive autoantibodies (AAB), is prone to costly diagnostic error. The relative importance of T cells to disease progression have long made it a target for prognostic markers, but the complexity of T cell sequencing data has hindered it's application in the field. In this study, we investigated the multiple types of TCR sequencing data as a source of predictive features for future onset of T1D. Using a large dataset of over 730 TCR repertoires from 321 patients, we found that TCR gene usage was highly predictive for clinical disease risk. Immunogenetic background played an important role in feature selection, stressing the importance of personalized medicine approaches to prognostic marker identification. We demonstrate that the same predictive features can be extracted from bulk RNA-seq data which is cheaper to measure than TCR profiling data. The features identified in this study can be used to measure disease progression in pre-clinical samples opening the door for future development of fast and accessible blood tests for T1D.