haplofreq: Fit Hardy-Weinberg equilibrium model for haplotype... In HaploSurvival: Modelling of haplo-effects for survival data, including competing risks. Effect of haplomatch for BMT studies.

Description

Fit MLE of Hardy-Weinberg equilibrium model for haplotype frequencies based on genotype data

Input is genotype data.

The haplotype model assumes Hardy-Weinberg equilibrium

P(H=(h_k,h_j)) = π_k pi_j

with K different haplotypes, where we parametrize the frequencies as

π_k= \frac{ \exp(θ_k) }{\exp(θ_1)+...+\exp(θ_{K-1})+1}

We allow a regression design on the haplotype parameters to reduce the dimensionality

θ = G α

where G is matrix of dimension (K-1) x (d-1). The α and θ version both have the same baseline that is a single haplotype.

Usage

 1 2 3 haplo.freqs(geno.type, geno.setup=NULL, Nit = 10,detail = 0, haplo.freq = NULL, step = 1,lev.marq=1,min.lev.marq=0, haplo.design=NULL,haplo.baseline=NULL,alpha=NULL) 

Arguments

 geno.type a data.frame with the variables. geno.setup setup form genotype data. Nit number of iterations for Newton-Raphson algorithm. detail if 0 no details is printed during iterations, if 1 details are given. haplo.freq haplo type frequencies for starting values or for fixed values. step step size for iteration steps. lev.marq starting value for Levenberg-Marquart algorithm. min.lev.marq stopping value for Levenberg-Marquart algorithm after initial steps, min.lev.marq=0 is stopping value for Levenberg-Marquart algorithm after initial steps, min.lev.marq=0 is Newton-Raphson steps, that are needed to get correct standard errors. haplo.design design for haplo parameters, default is diagonal of size K-1 haplo.baseline gives the name of the geno.setup that should be used as baseline, default is last haplotype. alpha starting value of parameters for haplo.design

Value

returns an object of type "haplo.freq". With the following arguments:

 haplo.alpha MLE of alpha parameters for haplo.design. haplo.pars MLE of haplotype parameters. haplo.freq MLE of haplotype frequencies. var.haplo.alpha variance matrix for alpha D2linv second derivative of log-likelihood. score score for likelihood. haplo.design design that relates haplotype parameters to alpha parameters. alpha.iid iid decomposition of alpha parameters, in this case score contributions for each subject for score of for log-likelihood. freq.names names of haplotypes

Author(s)

Thomas Scheike and Jeremy Silver

References

Scheike, Martinussen and Silver, Haplotype effects for survival data, Submitted.

Examples

  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 n<-200 g<-matrix(rbinom(200*4,1,0.5),n,4); # simple genotypes gs<-geno.setup(g); hgs<-haplo.freqs(g,geno.setup=gs) summary(hgs); hgs$haplo.freq # haplo-type frequencies hgs$haplo.pars # related parameters hgs$haplo.alpha # related parameters haplo.pars = X alpha hgs$var.haplo.alpha # variance of alpha parameters # structured haplo-frequency model, baseline is set to "0,0" gs<-geno.setup(g,haplo.baseline="0,0"); X<-matrix(1,3,1); hgs<-haplo.freqs(g,geno.setup=gs,haplo.design=X) summary(hgs); hgs$haplo.freq # haplo-type frequencies hgs$haplo.pars # related parameters hgs$haplo.alpha # related parameters haplo.pars = X alpha hgs$var.haplo.alpha # variance of alpha parameters 

HaploSurvival documentation built on May 2, 2019, 5:49 p.m.