Detection of homozygosity from next generation sequencing data using hidden markov model approach

Degrees of consanguinity has been a research tool since the establishment of medical genetics. Initial studies relied on large pedigree analyses using genetic analysis methods such as mapping of short tandem repeats. Long has been these methodologies were considered a gold standard for the discovery of long homozygous stretches called runs of homozygosity. These regions are considered to be the result of increased consanguinity and usually contain recessive deleterious disease causing mutations. With the ever increasing number of human genome projects all over the globe and advancements in high throughput genotyping methodologies and computation power, several new algorithms have been developed to detect runs of homozygosity within not only large families but also in large populations. In this study, we developed a simple, alternative strategy by integrating the allelic distributions within the X chromosome non-pseudoautosomal region to detect the runs of homozygosity from next generation sequencing data. Combining this new model with genotype probabilities within the genotype data inside our dynamic hidden Markov model algorithm, we generated a new tool namely ROHMM. It is implemented in java and contains both a command-line and a simple to use graphical user interface. ROHMM is tested on simulated data, real population data from 1000 Genome Project and clinical samples. Our results have shown that ROHMM can perform robustly producing highly accurate homozygosity estimations under all conditions thereby meeting and even exceeding the performance of both HMM based and sliding window based competitors.