plr: Firth's penalised logistic regression

In wanyuac/GeneMates: Analysing association and physical linkage between bacterial genes

Firth's penalised logistic regression

Description

Fit pairwise allelic presence/absence data with Firth's penalised logistic regression without a control for bacterial population structure. Because this function is not the focus of GeneMates, it takes as inputs the outputs of the lmm or findPhysLink function. As a result, users must run either function before calling this one.

Dependency: parallel, logistf and data.table

Usage

plr(pat, tests.pat, tests.allele, n.cores = -1, p.adj.method = "bonferroni")

Arguments

pat	An uncentred matrix of patterns of the allelic presence/absence status Expect every cell of this matrix to be a dichotomous variable.
tests.pat	A data frame specifying pairs of patterns whose association will be tested for with the logistic regression.
tests.allele	A data frame recording pairs of alleles whose association have been tested for by the findPhysLink function.
n.cores	Number of cores used to run GEMMA in parallel where possible. -1: automatically detect the number of available cores N, but use N - 1 cores (recommended) 0: automatically detect the number of available cores and use all of them. Be careful when the current R session is not running through SLURM. >= 1: use the number of cores as specified. n.cores is reset to the maximum number of available cores N when n.cores > N.
p.adj.method	Method for correcting p-values. Refer to the base function p.adjust for legitimate values of this argument.

Author(s)

Yu Wan ([e-mail])

Examples

assoc <- findPhysLink(...)
lr <- plr(pat = assoc$alleles$B, tests.pat = assoc$lmms.pat$dif$h1[, c("y_pat", "x_pat")],
      tests.allele = assoc$tests$dif$tests, n.cores = 8, p.adj.method = "bonferroni")

wanyuac/GeneMates documentation built on Aug. 12, 2022, 7:37 a.m.