The advances of high-throughput sequencing offer an unprecedented opportunity to study genetic variation. This is challenged by the difficulty of resolving variant calls in repetitive DNA regions. We present a Bayesian method to estimate repeat-length variation from paired-end sequence read data. The method makes variant calls based on deviations in sequence fragment sizes, allowing the analysis of repeats at lengths of relevance to a range of phenotypes. We demonstrate the method’s ability to detect and quantify changes in repeat lengths from short read genomic sequence data across genotypes. We use the method to estimate repeat variation among 12 strains of Arabidopsis thaliana and demonstrate experimentally that our method compares favourably against existing methods. Using this method, we have identified all repeats across the genome, which are likely to be polymorphic. In addition, our predicted polymorphic repeats also included the only known repeat expansion in A. thaliana, suggesting an ability to discover potential unstable repeats.

Triplet repeat expansions underlie several human genetic diseases such as Huntington's disease and Friedreich's ataxia. Although such mutations are primarily known from humans, a triplet expansion associated genetic defect has also been reported at the IIL1 locus in the Bur-0 accession of the model plant Arabidopsis thaliana. The IIL1 triplet expansion is an example of cryptic genetic variation as its phenotypic effects are seen only under genetic or environmental perturbation, with high temperatures resulting in a growth defect. Here we demonstrate that the IIL1 triplet expansion associated growth defect is not a general stress response and is specific to particular environmental perturbations. We also confirm and map genetic modifiers that suppress the effect of IIL1 triplet repeat expansion. By collecting and analyzing accessions from the island of Ireland, we recover the repeat expansion in wild populations suggesting that the repeat expansion has persisted at least 60 years in Ireland. Through genome-wide genotyping, we show that the repeat expansion is present in diverse Irish populations. Our findings indicate that even deleterious alleles can persist in populations if their effect is conditional. Our study demonstrates that analysis of groups of wild populations is a powerful tool for understanding the dynamics of cryptic genetic variation.

No abstract