Genetic Prediction and Screening of Early Life Polygenic Autoimmunity
Bibliographic record
Abstract
Autoimmune disorders occur when the immune system inappropriately reacts to self, causing the destruction of healthy tissue. Most polygenic autoimmune disorders share a common genetic architecture with strong associations in human leukocyte antigen (HLA) region genes which are critical to antigen specific immunity. The recent widespread availability of affordable single nucleotide polymorphism (SNP) array genotyping has enabled large case-control genome-wide association studies (GWAS) of polygenic traits. Polygenic risk scores (PRS) utilising SNPs identified from GWAS have enabled discrimination and prediction of polygenic traits. However, the unique architecture of polygenic autoimmunity requires special consideration and there exists scope for new methodologies to improve prediction of autoimmunity. In this thesis I focus on Type 1 Diabetes (T1D) and Coeliac Disease (CD) where existing PRS have demonstrated potential to classify diabetes type, predict incident cases and provide insight into research questions. Challenges in modelling HLA region associated risk arise due to its highly polymorphic nature, long range linkage disequilibrium (LD) and epistatic effects further confounded by nomenclature predating modern genetic understanding. In Chapter 2, focusing on T1D and using case-control data from the Type 1 Diabetes Genetics Consortium (T1DGC) I develop a new method to accurately mark molecular HLA haplotypes at the DR/DQ locus by studying linkage disequilibrium between SNPs and HLA haplotypes. Using T1DGC cohort genetic data I then develop an improved PRS model (“T1D-GRS2”) by accounting for haplotype interactions at the DR/DQ locus. By standardising the PRS to the UK Biobank population cohort I demonstrate potential thresholds for population screening and prediction. However, these results were developed and validated in European ancestry samples only, thus I was unable to demonstrate applicability to other ancestry populations. Further to this work, in Chapter 3 I applied the same methodology using CD case- control genotyping data. CD occurs due an autoimmune reaction to the ingestion of gluten peptides and current genetic testing employed in clinic does not factor in polygenic risk and is not cost-effective. In contrast to the previous chapter, we limited selection of SNPs for the “CD-GRS” PRS to those that we were able to develop into a direct genotyping panel and validated the PRS panel in a small clinical cohort from a paediatric gastroenterology clinic at Stollery Children’s Hospital, Alberta. By validating the PRS panel in a clinical setting and again standardising to the UK Biobank we were able to suggest potential use cases such as the PRS panel as an affordable rule-out test to avoid gastroscopy, which carries risks in a paediatric population. The analysis was again limited to European ancestry participants. Throughout my previous chapters and in ongoing studies of PRS there is a long- standing question over the applicability of PRS developed in European ancestry cohorts to non-European populations. In Chapter 4 I examine polygenic risk for T1D and T2D in a mixed ancestry US population youth with diabetes and examine the use of PRS in classification of diabetes type, and in prediction of the onset of T1D associated autoimmunity. I demonstrate the “T1D-GRS2” is broadly effective in a mixed ancestry population, however additional work is needed to more closely study subpopulations and determine if population specific PRS are needed or if trans- ancestry PRS can be broadly applied in all cases such as “T1D-GRS2”. Screening and follow-up studies are crucial to develop effective interventions such as immunotherapies which have demonstrated the ability to delay onset of T1D. In Chapter 5, tying together my previous findings I develop a combined screening panel for newborn screening of T1D and CD using a neural network model validated in Seattle area samples. The model and panel are made publicly available and easily reproducible for new and future screening studies. In this thesis I have improved upon the characterisation of polygenic risk associated with T1D and CD and expanded upon potential applications in clinical and research settings, particularly for population screening and intervention studies. I In Chapter 6 I discuss my findings, the implications of those findings, limitations and future steps and transferability to other polygenic autoimmune disorders.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".