Empirical affected-sib-pair statistics: Two simulation strategies

Abstract

We compare two simulation procedures used to determine empirical distributions of three affected sib-pair test-statistics in family samples with missing parental marker information. P1 and P2 procedures differ according to whether affected sibs' genotypes are ignored (P1) or accounted for (P2) to generate the unknown parental genotypes. Overall, our results show that P2 procedure performs better than P1, particularly when marker heterozygosity is high (H = 75%) or when marker data from unaffected siblings are available with low marker heterozygosity (H = 50%). These results are similar for all affected sib-pair tests.