Power to detect linkage in genomic screens using moving average test statistics

当使用连锁分析进行基因组分析时

When using linkage analysis to perform genomic screening for complex traits, regions may be identified where two-point linkage analysis produces modest p-values that, taken individually, would not be considered to be suggestive of linkage. One approach is to combine information from correlated clusters of markers into ?regional? tests of linkage. The properties of two classes of regional linkage test statistics were investigated over a range of heritabilities and different marker sets. The first class included minimum t values from sets of consecutive markers, the second included moving averages of the t-statistics. Initial results, based on a 10 cM map, suggest that the power for the min-t statistic, requiring that two consecutive t-tests be significant, was increased over two- point tests. Larger gains were observed for moving averages of 2 to 5 consecutive t-test values. The gains in power to detect linkage ranged between 5 and 20% for heritabilities between 30% and 60%. A novel ad hoc test proceedure analagous to a rank test was also considered, and its statistical properties are being investigated further.