gwrpv {gwrpvr}

Genome-Wide Regression P-Value (gwrpv) in R

Description

Computes the sample probability value (p-value) for the estimated coefficient from a standard genome-wide univariate regression. It computes the exact finite-sample p-value under the assumption that the measured phenotype (the dependent variable in the regression) has a known Bernoulli-normal mixture distribution.

Usage

gwrpv(beta, n0, n1, n2, mua, siga, mub, sigb, pa, pb, logdelta = -16,
  lognearnorm = -5, logtopsum = 8)

Arguments

beta

the beta being tested

n0

number of major allele homozygotes

n1

number of major allele heterozygotes

n2

number of minor allele zygotes

mua

parameter of the mixture distribution, can be any real number

siga

parameter of the mixture distribution, can be any real number

mub

parameter of the mixture distribution, can be any real number

sigb

parameter of the mixture distribution, can be any real number

pa

parameter of the mixture distribution, a real number between zero and one with pa+pb=1

pb

parameter of the mixture distribution, a real number between zero and one with pa+pb=1

logdelta

must be in log base 10 format, with default value set to -16

lognearnorm

must be in log base 10 format, with default value set to -5

logtopsum

must be in log base 10 format, with default value set to 8

Value

gwrpv returns a list containing:

$pvalue

p-value of a two-sided hypothesis test for a true coefficient of zero

$skew

skewness

$kurt

kurtosis of the coefficient estimate under assumed model

$skiptype

type of trimming/skip which took place (zero means no trimming)

$totnobs

total number of observations

$loopruns

number of sums in the main computation for each regression case

.

Examples

beta <- 6.05879
n0 <- 499
n1 <- 1
n2 <- 0
mua <- 13.87226
siga <- 2.58807
mub <- 4.62829
sigb <- 2.51803
pa <- 0.96544
pb <- 0.03456    # alternatively: pb <- 1.0 - pa
gwrpv(beta,n0,n1,n2,mua,siga,mub,sigb,pa,pb)

# note default values have been used for the trim parameters above
# in the following example we explicitly set the trim parameters
#
g <- gwrpv(beta,n0,n1,n2,mua,siga,mub,sigb,pa,pb,logdelta=-16,lognearnorm=-5,logtopsum=8)
g$pvalue