Detection of runs of homozygosity from whole exome sequencing data: state of the art and perspectives for clinical, population and epidemiological studies.

Runs of homozygosity (ROH) are sizeable stretches of homo- zygous genotypes at consecutive polymorphic DNA marker positions, traditionally captured by means of genome-wide single nucleotide polymorphism (SNP) genotyping. With the advent of next-generation sequencing (NGS) technologies, a number of methods initially devised for the analysis of SNP array data (those based on sliding-window algorithms such as PLINK or GERMLINE and graphical tools like Homozygosi- tyMapper) or specifically conceived for NGS data have been adopted for the detection of ROH from whole exome se- quencing (WES) data. In the latter group, algorithms for both graphical representation (AgileVariantMapper, HomSI) and computational detection (H3M2) of WES-derived ROH have been proposed. Here we examine these different approach- es and discuss available strategies to implement ROH detec- tion in WES analysis. Among sliding-window algorithms, PLINK appears to be well-suited for the detection of ROH, especially of the long ones. As a method specifically tailored for WES data, H3M2 outperforms existing algorithms espe- cially on short and medium ROH. We conclude that, notwith- standing the irregular distribution of exons, WES data can be used with some approximation for unbiased genome-wide analysis of ROH features, with promising applications to ho- mozygosity mapping of disease genes, comparative analysis of populations and epidemiological studies based on con- sanguinity.