Long range phasing has been a topic of interest since its inception by Kong (2008) [12] and has great potential for the analysis of genomic data, particularly as cohort sizes increase and hence more IBD sharing becomes present between individuals. Whilst the deCODE project has generated some excellent results, they have the advantage of an extremely powerful data set containing substantial amounts of IBD sharing which allows a rule based approach to long range phasing that yields very accurate haplotypes. This is not a luxury available to many research groups. We demonstrate that SHAPEIT2 implicitly performs this very accurate long range phasing when possible, whilst still leveraging LD when it is not.
