Skip to main content
. 2014 Apr 17;10(4):e1004234. doi: 10.1371/journal.pgen.1004234

Table 1. Switch error rates for samples containing nominally unrelated individuals.

Cohort CARL FVG GPC KOR ORC SPL VB VIS
Chip 370K 370K 2.5S 370K 300K 370K 370K 300K
N validation individuals 130 274 419 118 201 50 481 150
SLRP Yield 71.82 73.82 45.17 67.52 88.15 31.82 87.63 71.09
SHAPEIT2 (IBD) 0.75 0.21 0.16 0.18 0.14 0.60 0.17 0.19
SLRP (IBD) 1.15 0.40 0.28 0.45 0.33 1.99 0.33 0.43
Beagle (IBD) 2.57 1.18 0.30 1.25 0.94 4.38 1.07 1.30
HAPI-UR 3× (IBD) 5.43 2.25 0.35 2.55 2.59 8.30 1.93 2.94
SHAPEIT2 (No IBD) 5.03 4.10 0.51 3.11 2.43 3.35 2.74 3.65
Beagle (No IBD) 7.73 5.72 0.84 5.55 5.05 6.00 5.03 6.36
HAPI-UR 3× (No IBD) 15.97 10.31 0.82 10.20 9.94 12.09 7.92 12.47
SHAPEIT2 2.85 1.81 0.28 1.05 0.49 2.65 0.62 1.16
Beagle 5.31 3.11 0.49 2.53 1.78 5.57 1.88 2.78
HAPI-UR 3× 11.21 5.83 0.50 4.82 4.20 11.10 3.21 5.77

All individuals not explicitly related in the defined pedigrees were phased. We calculate overall SE rate (All), for chromosome 10 (this is not possible for SLRP) as well as SE rates within (IBD) and outside (no IBD) SLRP detected IBD regions. SHAPEIT2 consistently produces the most accurate haplotypes. Chip abbreviations: 370K - Illumina HumanHap 370CNV. 300K - Illumina HumanHap 300, 2.5S - Illumina Omni 2.5S.