Abstract
PURPOSE: The variome of the Turkish (TK) population, a population with a considerable history of admixture and consanguinity, has not been deeply investigated for insights on the genomic architecture of disease. METHODS: We generated and analyzed a database of variants derived from exome sequencing data of 773 TK unrelated, clinically affected individuals with various suspected Mendelian disease traits and 643 unaffected relatives. RESULTS: Using uniform manifold approximation and projection, we showed that the TK genomes are more similar to those of Europeans and consist of 2 main subpopulations: clusters 1 and 2 (N = 235 and 1181, respectively), which differ in admixture proportion and variome (https://turkishvariomedb.shinyapps.io/tvdb/). Furthermore, the higher inbreeding coefficient values observed in the TK affected compared with unaffected individuals correlated with a larger median span of long-sized (>2.64 Mb) runs of homozygosity (ROH) regions (P value = 2.09e-18). We show that long-sized ROHs are more likely to be formed on recently configured haplotypes enriched for rare homozygous deleterious variants in the TK affected compared with TK unaffected individuals (P value = 3.35e-11). Analysis of genotype-phenotype correlations reveals that genes with rare homozygous deleterious variants in long-sized ROHs provide the most comprehensive set of molecular diagnoses for the observed disease traits with a systematic quantitative analysis of Human Phenotype Ontology terms. CONCLUSION: Our findings support the notion that novel rare variants on newly configured haplotypes arising within the recent past generations of a family or clan contribute significantly to recessive disease traits in the TK population.