Table 3.

Population-based haplotype estimation tools used in this study

Tool-Version	Year	Heuristic to reduce haplotype search space
fastPHASE - 1.4.8	2006	Uses a haplotype-clustering model, where the set of all possible haplotypes are clustered into a small fixed number of “ancestral” haplotypes [31].
BEAGLE - 4.1	2017	Uses a haplotype-clustering model with a variable number of clusters, depending on the region under consideration [24].
IMPUTE - 2.3.2	2009	Subsamples possible haplotypes that are similar to those of the currently estimated haplotype of an individual [32].
MaCH - 1.0.18.c	2010	Subsamples possible haplotypes from the set of all possibilities randomly at each iteration [33].
SHAPEIT2 - v2.r837	2012	Breaks the chromosome into small windows of a few SNPs, estimates the phase of each window using a method similar to IMPUTEv2 and then estimates transitions between windows [23].
HAPI-UR - v1.01	2012	Breaks chromosome into small windows, that are initially very short but iteratively grow to a user defined size, enabling modelling of longer segments at once [34].
EAGLE2 - v2.3.5	2016	When no reference panel is provided (the scenario in this study), EAGLE2 applies long range phasing (EAGLE1 [27]) then efficiently represents all haplotypes such that beam search can be applied to evaluate only the most promising phase paths [22].