histo_logit_cov: perform logistic regression for a binary/binarized trait with...

In SingerLab/gac: Genetic Analysis of Cells

perform logistic regression for a binary/binarized trait with a covaritate

Description

These functions were developed for the genetic analysis of cancer subtypes, where "histology" is considered the phenotype, and copy number the genotype. The association is carried out using logistic regresion by 'glm' using 'family = "binary"', as this recreates the standard model of quantitative genetics P = G + E. Alternate arguments of 'family' have no been tested.

Usage

histo_logit_cov(
  cnr,
  trait,
  pheno0,
  pheno1,
  covar,
  exclude.cluster = "HC",
  family = "binomial",
  na.action = "na.exclude",
  ...
)

Arguments

cnr	a cnr bundle
trait	character, name of the trait of interest to analyze. Must be a column in the phenotype matrix (Y). e.g. "binary1"
pheno0	character, phenotype(s) to use as baseline, e.g. "0"
pheno1	character, phenotype(s) to use as alternate, e.g. "1"
covar	character, model covariates to include in the model, e.g. "category1"
exclude.cluster	character, list of clusters to exclude, e.g. hypersegmented, Stroma, etc. Default "HC"
family	character, description of the error distribution and link function to be used in the model. See glm for details. Default "binomial"
na.action	character, handling of NA. default is "na.exclude"
...	additional arguments passed to glm

Value

a CNR object with results from a logistic regression analysis (family = "binomial") with effect estimates, and p-values attached to the chromInfo and gene.index matrices.

Results columns are "Estimate", "Std.Error", "z.value", "p.value", and "q.value"; with the phenotype comparison pre-apended as <pheno0>.vs.<pheno1>.cv<covar>.lr.<value>. Using grade as an example the columns would be 0.vs.1..lr.Estimate, 0.vs.1.quantitative1.lr.Std.Error, 0.vs.1.quantitative1.lr.z.value, 0.vs.1.quantitative1.lr.p.value, and 0.vs.1.quantitative1.lr.q.value

Examples

data(cnr)

cnr <- histo_logit_cov(cnr, trait = "binary1",
   pheno0 = 0, pheno1 = 1, covar = "category1")

cnr <- histo_logit_cov(cnr, trait = "category1",
   pheno0 = "A", pheno1 = c("B", "C"), covar = "category2")

cnr <- histo_logit_cov(cnr, trait = "category2",
   pheno0 = c("X", "Y"), pheno1 = "Z", covar = "category1")

SingerLab/gac documentation built on March 23, 2024, 5:15 a.m.