Averaged over the two imputation datasets, correlation coefficients between imputed and true genotypes for autosomal markers, pseudo-autosomal markers, and X-specific markers were 0.971, 0.831 and 0.935 when using Findhap, and 0.983, 0.856 and 0.937 when using Beagle.
Histograms were generated to show the distribution of the INFO scores for each bin and the distribution of the differences in INFO scores, and we estimated the correlation between the INFO scores for the two imputation approaches.
In the five imputation data sets the EPV varied between 19.1 and 19.6 for disability and between 13.5 and 13.8 for pain intensity.
In this multiple imputation procedure, standard errors were decomposed to the variability across and within the five imputations (Enders [2010]; OECD [2005]; Wu [2005]).
To provide an overall comparison between the three applied imputation methods, we set this threshold to 0, such that all genotypes were called.
For imputation of CA, the results were similar between the three methods for imputation, with Scenarios 1 and 2 producing the best concordance and Scenarios 1 and 3 producing the highest mean and minimum quality score.
Imputation between the two recombination hotspots encompassing each of the 24 loci that showed evidence of association in Phase I was accomplished with Impute v2 using two reference panels: 1000 Genomes Project (b36) for wide coverage, and HapMap3 (r2 b36) for deep coverage [ 16, 20, 21].
The expected difference E[ S2 − S1] in imputation accuracy between the two panels is also affected by exponential growth.
In the event that patients are missing baseline characteristics data, which are otherwise imbalanced between the two study groups, we will use imputations to estimate models which include those characteristics.
