run.bvs: Bayesian variable selection

In chr1swallace/GUESSFM: Using GUESS for Fine Mapping

Description

Wrapper for GUESS, an alternative to using R2GUESS

Usage

run.bvs(X, Y, gdir = "test", sub = NA, covars = NULL,
  family = "gaussian", nsweep = 55000, nchains = 3, nexp = 3,
  tag.r2 = 0.99, nsave = 1000, guess.command = NULL, wait = FALSE,
  run = TRUE, ...)

Arguments

X	snpMatrix object, holding genotypes
Y	vector or matrix of phenotypes, nrow(Y)==nrow(X)
gdir	directory where all the GUESS input/output files will be. If it doesn't exist, it will be created.
sub	optional number < nrow(X). If supplied, only the subset of samples defined by 1:sub will be used.
covars	optional matrix or vector of covariates. If supplied GUESS will be run on residuals from glm(Y ~ ., data=as.data.frame(covars)).
family	family for Y ~ covars regression. default "gaussian".
nsweep	number of sweeps, see documentation for GUESS
nchains	number of chains, see documentation for GUESS
nexp	expected number of causal variants in region
tag.r2	r squared value at which to tag to avoid numerical instability. Default of 0.99 has worked well in our experience.
nsave	number of models to save, see documentation for GUESS
guess.command	Command to run GUESS. This is normally automatically set to the version of GUESS installed by R2GUESS, but you may override with a full path to a system version of GUESS if you prefer.
wait	logical. default FALSE. if TRUE, run.bvs will wait for GUESS to finish, rather than running in background
run	logical. default TRUE. If FALSE, don't run GUESS, just return the command that would be run
...	GUESS starts from a stepwise solution found by cond.best. Use ... to pass arguments firectly to cond.best to influence the p value threshold or number of predictors at which the stepwise search stops.

Value

nothing. side effect is to set GUESS running in the background. This takes a while (typically several hours).

chr1swallace/GUESSFM documentation built on May 13, 2019, 6:17 p.m.