haplotypeOddsRatio {genepi}

Calculate haplotype disease risk.

Description

Haplotype disease risk is calculated resolving haplotype ambiguity and adjusting for covariates and population stratification.

Usage

haplotypeOddsRatio(formula, gtypevar, data, stratvar=NULL, nsim=100, tol=1e-8)
## S3 method for class 'haploOR'
print(x, ...)

Arguments

formula

The formula for logistic regression without the haplotype variable.

gtypevar

The variable names in the data frame corresponding to the loci of interest. Each variables counts the number of mutant genotypes a subject has at that locus. Legal values are 0, 1, 2 & NA.

data

The name of the dataframe being analyzed. It should have all the variables in the formula as well as those in genotype and stratvar.

stratvar

Name of the stratification variable. This is used to account for population stratification. The haplotype frequencies are estimated within each stratum.

nsim

Variance should be inflated to account for inferred ambiguous haplotypes. The estimates are recalculated by simulating the disease haplotype copy number and variance added to average.

tol

Tolerance limit for the EM algorithm convergence.

x

Object of class haploOR.

...

Other print options.

Details

This implements the method in the reference below.

Value

It is a list of class haploOR

call

The function call that produced this output.

coef

Table with estimated coefficients, standard error, Z-statistic and p-value.

var

Covariance matrix of the estimated log odds-ratiios.

deviance

Average of the simulated deviances. Its theoretical properties are unknown.

aic

Average of the simulated aic.

null.deviance

Deviance of null model.

df.null

Degrees of freedom of null model.

df.residual

Degrees of freedom of full model.

The "print" method formats the results into a user-friendly table.

Author(s)

Venkatraman E. Seshan

References

Venkatraman ES, Mitra N, Begg CB. (2004) A method for evaluating the impact of individual haplotypes on disease incidence in molecular epidemiology studies. Stat Appl Genet Mol Biol. v3:Article27.

Examples

# simulated data with 2 loci haplotypes 1=00, 2=01, 3=10, 4=11
# control haplotype probabilities p[i]  i=1,2,3,4
# haplotype pairs (i<=j) i=j: probs = p[i]^2 ; i<j: p[i]*p[j]
p <- c(0.25, 0.2, 0.2, 0.35)
p0 <- rep(0, 10)
l <- 0
for(i in 1:4) {for(j in i:4) {l <- l+1; p0[l] <- 2*p[i]*p[j]/(1+1*(i==j))}}
controls <- as.numeric(cut(runif(1000), cumsum(c(0,p0)), labels=1:10))
# case probabilities disease haplotype is 11
or <- c(2, 5)
p1 <- p0*c(1,1,1,2,1,1,2,1,2,8); p1 <- p1/sum(p1)
cases <- as.numeric(cut(runif(1000), cumsum(c(0,p1)), labels=1:10))
# now pool them together and set up the data frame
dat <- data.frame(status=rep(0:1, c(1000, 1000)))
# number of copies of mutant variant for locus 1
dat$gtype1 <- c(0,0,1,1,0,1,1,2,2,2)[c(controls, cases)]
# number of copies of mutant variant for locus 2
dat$gtype2 <- c(0,1,0,1,2,1,2,0,1,2)[c(controls, cases)]
# true number of copies of disease haplotype
dat$hcn <- c(0,0,0,1,0,0,1,0,1,2)[c(controls, cases)]
# model with genotypes only
haplotypeOddsRatio(status ~ 1, c("gtype1","gtype2"), dat)
# model from the logistic fit using the number of copies of disease haplotype
glm(status ~ factor(hcn), dat, family=binomial)