Local false discovery rate and minimum total error rate approaches to identifying interesting chromosomal regions

Abstract

The simultaneous testing of a large number of hypotheses in a genome scan, using individual thresholds for significance, inherently leads to inflated genome-wide false positive rates. There exist various approaches to approximating the correct genomewide p-values under various assumptions, either by way of asymptotics or simulations. We explore a philosophically different criterion, recently proposed in the literature, which controls the false discovery rate. The test statistics are assumed to arise from a mixture of distributions under the null and non-null hypotheses. We fit the mixture distribution using both a nonparametric approach and commingling analysis, and then apply the local false discovery rate to select cut-off points for regions to be declared interesting. Another criterion, the minimum total error, is also explored. Both criteria seem to be sensible alternatives to controlling the classical type I and type II error rates.

Comparison of single-nucleotide polymorphisms and microsatellite markers for linkage analysis in the COGA and simulated data sets for Genetic Analysis Workshop 14: Presentation Groups 1, 2, and 3

“…This data set included four populations: Aiputo (AI), Danaca (DA), Karanga (KA), and New York (NY), with 100 replicates from each population. Eight groups analyzed the AI, DA, and KA data [Babron et al, 2005;Klein et al, 2005;Lin and Liu, 2005;Papachristou and Lin, 2005;Sinha et al, 2005;Thalamuthu et al, 2005;Wu et al, 2005;Yue et al, 2005]. One of these groups analyzed these three populations, but focused their reported results on the DA population [Thalamuthu et al, 2005].…”

“…Eight groups using the simulated data requested the answers prior to analysis [Ghosh et al, 2005;Klein et al, 2005;Papachristou and Lin, 2005;Sinha et al, 2005;Thalamuthu et al, 2005;Wu et al, 2005;Xing et al, 2005;Yang et al, 2005]. Four groups [Babron et al, 2005;Lin and Liu, 2005;Plancoulaine et al, 2005;Yue et al, 2005] did not request the answers.…”

“…Most of the groups analyzing the simulated data used the Kofendrerd personality disorder (KPD) trait [Babron et al, 2005;Lin and Liu, 2005;Papachristou and Lin, 2005;Plancoulaine et al, 2005;Sinha et al, 2005;Thalamuthu et al, 2005;Wu et al, 2005;Xing et al, 2005;Yang et al, 2005b;Yue et al, 2005]. One group used logistic regression to identify five personality traits that were significantly associated with KPD [Ghosh et al, 2005].…”

“…Six of these eight groups analyzed all 100 replicates. The remaining two groups combined data from subsets of replicates: Sinha et al [2005] analyzed either 10 or 20 randomly chosen replicates and merged replicates to form samples of 200 families and 400 families, while Lin and Liu [2005] pooled the last three replicates. Two groups analyzed data from only one of these three populations: Xing et al [2005] randomly chose one replicate of the KA nuclear pedigrees, while Plancoulaine et al [2005] combined the data in replicates 1 and 2 of the AI population.…”

Prediction of Empirical p Values from Asymptotic p Values for Conditional Logistic Affected Relative Pair Linkage Analysis

“…The multiple testing issue becomes signifi cant when a large number of tests is performed with different covariates. Bonferroni's correction and several other methods to allow for multiple testing are available in the literature to correct p values for multiple tests [1][2][3] . In addition to the multiple testing problem, it is diffi cult to have confi dence in the very small asymptotic p values often obtained, especially when covariates are included in the models used for linkage analysis.…”