The genotypes for all of the autosomal markers were analyzed for each family by using Pedigree Relationship Statistical Test (PREST) [see McPeek and Sun, 2000] to detect sample and pedigree structure errors. DNA was re-isolated from a stored frozen blood specimen, and the genotyping was repeated for any individual for whom PREST detected a probable error. If regenotyping failed to resolve the error, the problematic genotype was subsequently treated as missing. The program Pedcheck was used to detect non-Mendelian inheritance [O'Connell et al., 1998]. Markers with a high frequency of Mendelian segregation errors were excluded from analysis, and for isolated Mendelian errors, the genotypes for the entire family were excluded for the specific marker that yielded the error. To further reduce errors, the probability that each genotype was correct was assessed by using the error-checking algorithm implemented in Merlin [Abecasis et al., 2002], in which genotypes that had a probability of less than 0.025 of being correct were removed from further consideration.
