True and false positive peaks in genomewide scans: The long and the short of it

When performing a genome scan in linkage or linkage disequilibrium studies to detect loci underlying complex or quantitative traits, it is important t o attempt to distinguish between true and false positives using the appropr iate statistical methods. There has been some controversy in the literature regarding the use of the length of a positive peak, i.e., the length of a chromosome region displaying identity-by-descent in linkage studies among a ffected individuals or the length of a continuous chromosome region for whi ch the test statistic is above a certain threshold. We show in this study, by reasoning and by simulation studies, that conditional on the strength of evidence for a locus affecting a trait of interest, i.e.. conditional on t he peak height of a test statistic, there is no information in the length o f the peak. Our finding has implications for linkage and association studie s.