shareTest {SHARE} | R Documentation |
Permutation tests to compute the experimentwise p-values that account for model searching.
shareTest(outObj, haploObj, status, tol = 1e-08, verbose = FALSE, nperm = 1000)
outObj |
the share object outputed from
cshare function |
haploObj |
The haplo object cshare applied to |
status |
A character string indicating the column name of the phenotype in
haploObj@pheno to be used as the clinical status in the analysis. |
tol |
The convergence parameter for the haplotype logistic regression. |
verbose |
TRUE/FALSE to decide whether to create log file for debug |
nperm |
maximal number of permutation tests |
If the best model size is zero, there appears to be no genetic association in the region of interest. There is no need to perform a permutation test. For final models with at least 1 SNPs, we permute case-control labels 1000 times regardless of the genotypic data, carry out model searching for each permuted dataset, and compute the nominal p-value using a Wald test. Finally the experiment-wise p-value is computed by comparing the observed p-value to its null distribution.
The experiment-wise p-value from the permutation test will be returned.
James Y. Dai
J. Y. Dai, M. LeBlanc, N. L. Smith, B. M. Psaty, and C. Kooperberg. SHARE: an adaptive algorithm to select the most informative set of SNPs for genetic association. Biostatistics, 2009. In press.
J. Besag and P. Clifford. Sequential monte carlo p-values. Biometrika, 78(2):301, June 1, 1991.
## Not run: ## See vignette for more details permuPValue <- shareTest(outObj=kerem[["Cross-Val"]], haploObj=keremHaplo, status = "CF", nperm=1000 ) ## End(Not run)