Premium
Detecting linkage for a complex disease using simulated extended pedigrees
Author(s) -
DeStefano Anita L.,
Cupples L. Adrienne,
Myers Richard H.,
Farrer Lindsay A.
Publication year - 1997
Publication title -
genetic epidemiology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.301
H-Index - 98
eISSN - 1098-2272
pISSN - 0741-0395
DOI - 10.1002/(sici)1098-2272(1997)14:6<981::aid-gepi70>3.0.co;2-g
Subject(s) - replicate , genome scan , false positive paradox , quantitative trait locus , pedigree chart , linkage (software) , statistics , regression , selection (genetic algorithm) , biology , data set , set (abstract data type) , regression analysis , trait , genetics , genetic linkage , identification (biology) , false positives and false negatives , mathematics , gene , computer science , artificial intelligence , microsatellite , allele , programming language , botany
The simulated extended pedigree data of the Genetic Analysis Workshop 10 were used to examine the relationship between several quantitative traits (Q1‐Q5), an environmental factor, age and sex and to identify genes contributing to the quantitative traits. A forward selection procedure was used to identify regression models for each trait. Residuals from these regression models were used as quantitative traits in linkage analysis. Two‐point sib‐pair analysis was performed on Replicate 1 of the data set using SIBPAL. Sixteen regions on 8 chromosomes yielded two‐point p‐values < 0.005 in Replicate 1. Two strategies for utilizing a second data set were evaluated. In a two‐stage approach, only those regions with p‐value < 0.005 in Replicate 1 were followed up in the second data set. Nine of these regions had p‐values < 0.05 in Replicate 2; four were associated with major genes included in the generating model and the remaining five regions were false positives. An alternative strategy was to perform a repeat genome wide screen in the second data set. This strategy resulted in the identification of 20 regions with p‐values < 0.05 in both replicates; five of which included major genes included in the generating model. Although the false positive rate increased when a complete genome screen was performed on both data sets, the two‐stage screen, with a more stringent initial criterion for identifying suggestive linkages, had a higher rate of false negatives. For some studies, conducting two complete genome screens in a split‐sample design may be worthwhile. © 1997 Wiley‐Liss, Inc.