| gJLS2s {gJLS2} | R Documentation |
generalized Joint-Location-Scale (gJLS) test with summary statistics
Description
This function takes as input the gL and gS p-values for each SNP and combine to produce the gJLS p-values. It is used for genome-wide analysis where only the gL or gS p-values are available, caution should be exercised when combing gL and gS p-values obtained from separate datasets.
Usage
gJLS2s(gL, gS)
Arguments
gL |
a vector of location p-values or a data.frame containing column names "SNP" and "gL". |
gS |
a vector of scale p-values or a data.frame containing column names "SNP" and "gS". |
Value
a vector of combined gJLS p-values for each SNP.
Note
For a genome-scan, we recommend to run this in PLINK via the plugin function gJLSPLINK, especially for large datasets and those with more than 20 covariates.
We highly recommend to quantile-normally transform Y for non-symmetrically distributed traits. This is typically done to avoid ‘scale-effect’ when the variance values tend to be proportional to mean values when stratified by GENO, as observed by Pare et al. (2010) and Yang et al. (2011).
For the moment, only quantitative trait Y is accepted as the subsequent generalized joint location scale (gJLS) analyses require the variance be calculated on quantitative traits. However, we are working on to include binary response for the generalized JLS analyses in the next update of gJLS.
Author(s)
Wei Q. Deng deng@utstat.toronto.edu, Lei Sun sun@utstat.toronto.edu
References
Soave D, Corvol H, Panjwani N, Gong J, Li W, Boëlle PY, Durie PR, Paterson AD, Rommens JM, Strug LJ, Sun L. (2015). A Joint Location-Scale Test Improves Power to Detect Associated SNPs, Gene Sets, and Pathways. American Journal of Human Genetics. 2015 Jul 2;97(1):125-38. doi: 10.1016/j.ajhg.2015.05.015. PMID: 26140448; PMCID: PMC4572492.
Examples
gL <- data.frame("SNP" = paste("rs", 1:100, sep=""), "gL"=runif(100))
gS <- runif(100)
gJLS2s(gL = gL, gS=gS)