Comparison of marker intervals and number of sib pairs used for linkage analysis on simulated nuclear family data.

Using a two-stage global scan design, we analyzed general population replicates 1 and 42 of the Genetic Analysis Workshop (GAW) 12 simulated data set using three methods: revisited Haseman-Elston (HER), maximum likelihood variance estimation (ML), and variance components (VC). Three marker densities, 5-, 10-, and 15-cM intervals, were examined in the first-stage scan. We found that the 10-cM interval appears to be the most cost-effective approach in genotyping without sacrificing power when using a first stage significance level of 0.01. Subsequently, we performed the second-stage scan at 1-cM intervals for those putative positive regions identified in the first-stage scan at a significance level of 0.01. We also compared the power to detect linkage using different numbers of sib pairs for a genome-wide scan at a 10-cM interval and found that power decreases nonlinearly as the number of sib pairs decreases.