Premium
Genome Scans for Q1 and Q2 on General Population Replicates Using Loki
Author(s) -
Shmulewitz Dvora,
Heath Simon C.
Publication year - 2001
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/gepi.2001.21.s1.s686
Subject(s) - replicate , genome scan , linkage (software) , chromosome , population , genome , data set , genetics , biology , markov chain monte carlo , markov chain , set (abstract data type) , sampling (signal processing) , quantitative trait locus , computational biology , monte carlo method , computer science , statistics , mathematics , microsatellite , gene , allele , demography , filter (signal processing) , sociology , computer vision , programming language
The Markov Chain Monte Carlo linkage package Loki was used to perform a genome scan under realistic conditions (using a 10‐cM marker map without marker data on unsampled individuals, analyzing each chromosome separately, and without knowing the answers) for traits Q1 and Q2 on general population replicate 1. Using this approach we detected and correctly localized MG1 for Q1 and MG3 for Q2. We then repeated the analyses on replicate 1 and the “best replicate” (42) adding more information (using marker data on everyone, fitting a polygenic effect, and analyzing multiple chromosomes jointly) to see the effect on the detection of trait loci. We found that adding more data often improves the quality of the linkage signal, and reduces the false positive rate, but did not allow the detection of trait loci missed by the initial analysis. We also investigated the convergence of the sampler by repeating one multi‐chromosome analysis six times with different random number seeds. We concluded that a strategy of performing a single chromosome scan using a moderate number of sampling iterations, followed by a multi‐chromosome analysis of all chromosomes with linkage signals detected in the first scan using a longer sampling run, was an effective way of performing a genome scan on this data set. © 2001 Wiley‐Liss, Inc.