Abstract
Studying the genetics of disease susceptibility in populations that are small or understudied will further the understanding of the aetiology of rare and/or complex diseases. Biliary atresia is a rare, usually sporadic, fibrotic obliteration of the biliary tree that without timely treatment leads to liver failure in infancy, and is the indication for more than half of all paediatric liver transplant referrals worldwide. In Māori and Polynesian populations the incidence is elevated three-fold compared to Europeans. An iwi has been identified exhibiting an extremely elevated incidence, with about 1 in 100 children diagnosed within an established pedigree in the iwi. Methods were developed for this thesis to enrich for shared regions of the genome in affected individuals and hypothesised carriers. These methods collate loci where alleles are continuously shared across informative markers (i.e. are Identical-By-State (IBS)).
To understand the genetic background on which this susceptibility has arisen, the genetics of the patients compared to Māori samples outside the iwi was investigated. The rate of BA in the iwi and New Zealand was found to be unequivocally exceptional; BA occurs in the pedigree at 50x the rate in the general Māori population (P < 0.001) and is also significantly higher in patients outside the pedigree but in the iwi (P = 0.048). Multiple methods were used to ascertain an absence of evidence for excessive homozygosity in patients, suggesting that the susceptibility either acts dominantly or is polygenic. Genomic measures of sharing and variation between samples established that patients are part of an admixed Māori population exhibiting very high linkage disequilibrium but no evidence of iwi-dependent population iv structure. This suggests a history of substantial, historical, population bottlenecking and ubiquitous inter-iwi gene flow. This makes it unlikely that recessive or multiple additive low-risk alleles could have increased in allele frequency dramatically enough to account for the 50x increase in the rate of BA. A dominant factor, either at one locus with low penetrance or at a very small number of loci, was proposed as the model of BA in this Iwi.
To identify candidate loci for a disease with this unusual model, methods were developed to enrich for shared regions of the genome between affected individuals, and Next Generation Sequencing (NGS) was used to directly identify rare or new variants as candidates for the causal factor. These methods with NGS were applied to a disease unrelated to BA, Spondylocarpotarsal Syndrome (SCTS). This disease was hypothesised to be recessive through unreported relatedness between parents and families, as it was found in a group of families of unknown relatedness but who shared a surname, geographical proximity and camptodactyly, which is an unusual feature for this disease. This failed to resolve a locus and showed that there was no evidence for the hypothesised consanguinity, or relatedness between families. The tools were then optimised to identify a hypothesised dominant, BA susceptibility model on a group of iwi patients with known relatedness and a complex, admixed population history.
Next Generation Sequencing was utilised to identify shared variants across iwi patients with BA. Variants were filtered against a hypothesis of a conserved, rare, coding variant private to Māori present in all affected members of the pedigree and most of the larger iwi cohort. 15 exonic candidates were found, but because none fitted all aspects of the hypothesis the candidates pool was extended to include intergenic variants identified with whole genome sequencing (WGS). Detection of Copy Number Variants did not yield any candidates but is yet to be comprehensively excluded.
To substantiate the candidates from NGS and localise the search space of the genome, non-parametric analysis was performed iteratively across patient pairs to find peaks of IBS v runs of phased, informative markers optimised from Single Nucleotide Polymorphism (SNP) array and WGS. This provided evidence for several candidate loci. One peak in particular on chromosome 9, situated over the TLR4 gene, was consistent over multiple sharing hypotheses and fits the reported pedigree. Some peaks contained intergenic variants where further testing for involvement in the BA susceptibility is warranted, such as two highly conserved variants on chromosome 16. The method is applicable to other familial diseases where the model of inheritance, number of phenocopies, relatedness, consanguinity or allele frequencies are unknown.