Skip to main content
. 2007 Sep 21;81(5):1084–1097. doi: 10.1086/521987

Table 2. .

Allelic-Imputation Error Rates

Error Rate(%)
Low-Density Simulateda
High-Density Simulateda
Affymetrix 500K WTCCCb
Method Small Medium Large Small Medium Large Small Medium Large
Beaglec:
R=25 3.45 1.91 1.57 1.17 .53 .13 1.21 .87 .77
R=4 3.65 2.01 1.57 1.36 .59 .13 1.32 .88 .78
R=1 4.10 2.21 1.80 1.82 .70 .16 1.44 .94 .84
fastPHASEd:
K=20 3.37 2.54 2.80 1.19 .90 .57 1.12 .95 .93
K=10 3.91 3.19 3.43 1.72 1.38 1.04 1.21 1.05 1.04
HAP 5.26 3.66 3.89 3.20 2.55 2.31 1.68 1.38 1.33
2SNP 8.52 5.76 5.14 3.61 2.33 1.26 2.50 1.84 1.71
a

Parameters for selection of tag SNPs for 1 Mb of simulated data were chosen to obtain approximate densities of 1 SNP per 10 kb (low density) or 1 SNP per 3 kb (high density). For each density, three sample sizes were considered: 100 (small), 1,000 (medium), and 5,000 (large) individuals.

b

For the WTCCC data, sets of 200 consecutive markers were used. Three sample sizes were considered: 100 (small), 1,000 (medium), and 3,002 (large) individuals.

c

Beagle results are given for R=1, R=4, and R=25 samples per individual.

d

fastPHASE results are given for K=10 and K=20 clusters per individual.