mets (version 1.3.4)

haplo.surv.discrete: Discrete time to event haplo type analysis

Description

Can be used for logistic regression when time variable is "1" for all id.

Usage

haplo.surv.discrete(
  X = NULL,
  y = "y",
  time.name = "time",
  Haplos = NULL,
  id = "id",
  desnames = NULL,
  designfunc = NULL,
  beta = NULL,
  no.opt = FALSE,
  method = "NR",
  stderr = TRUE,
  designMatrix = NULL,
  response = NULL,
  idhap = NULL,
  design.only = FALSE,
  covnames = NULL,
  fam = binomial,
  weights = NULL,
  offsets = NULL,
  idhapweights = NULL,
  ...
)

Arguments

X

design matrix data-frame (sorted after id and time variable) with id time response and desnames

y

name of response (binary response with logistic link) from X

time.name

to sort after time for X

Haplos

(data.frame with id, haplo1, haplo2 (haplotypes (h)) and p=P(h|G)) haplotypes given as factor.

id

name of id variale from X

desnames

names for design matrix

designfunc

function that computes design given haplotypes h=(h1,h2) x(h)

beta

starting values

no.opt

optimization TRUE/FALSE

method

NR, nlm

stderr

to return only estimate

designMatrix

gives response and designMatrix directly not implemented (mush contain: p, id, idhap)

response

gives response and design directly designMatrix not implemented

idhap

name of id-hap variable to specify different haplotypes for different id

design.only

to return only design matrices for haplo-type analyses.

covnames

names of covariates to extract from object for regression

fam

family of models, now binomial default and only option

weights

weights following id for GLM

offsets

following id for GLM

idhapweights

weights following id-hap for GLM (WIP)

...

Additional arguments to lower level funtions lava::NR optimizer or nlm

Author

Thomas Scheike

Details

Cycle-specific logistic regression of haplo-type effects with known haplo-type probabilities. Given observed genotype G and unobserved haplotypes H we here mix out over the possible haplotypes using that P(H|G) is provided.

$$ S(t|x,G)) = E( S(t|x,H) | G) = \sum_{h \in G} P(h|G) S(t|z,h) $$ so survival can be computed by mixing out over possible h given g.

Survival is based on logistic regression for the discrete hazard function of the form $$ logit(P(T=t| T \geq t, x,h)) = \alpha_t + x(h) \beta $$ where x(h) is a regression design of x and haplotypes \(h=(h_1,h_2)\)

Likelihood is maximized and standard errors assumes that P(H|G) is known.

The design over the possible haplotypes is constructed by merging X with Haplos and can be viewed by design.only=TRUE

Examples

## some haplotypes of interest
types <- c("DCGCGCTCACG","DTCCGCTGACG","ITCAGTTGACG","ITCCGCTGAGG")

## some haplotypes frequencies for simulations 
data(hapfreqs)

www <-which(hapfreqs$haplotype %in% types)
hapfreqs$freq[www]

baseline=hapfreqs$haplotype[9]
baseline

designftypes <- function(x,sm=0) {# {{{
hap1=x[1]
hap2=x[2]
if (sm==0) y <- 1*( (hap1==types) | (hap2==types))
if (sm==1) y <- 1*(hap1==types) + 1*(hap2==types)
return(y)
}# }}}

tcoef=c(-1.93110204,-0.47531630,-0.04118204,-1.57872602,-0.22176426,-0.13836416,
0.88830288,0.60756224,0.39802821,0.32706859)

data(hHaplos)
data(haploX)

haploX$time <- haploX$times
Xdes <- model.matrix(~factor(time),haploX)
colnames(Xdes) <- paste("X",1:ncol(Xdes),sep="")
X <- dkeep(haploX,~id+y+time)
X <- cbind(X,Xdes)
Haplos <- dkeep(ghaplos,~id+"haplo*"+p)
desnames=paste("X",1:6,sep="")   # six X's related to 6 cycles 
out <- haplo.surv.discrete(X=X,y="y",time.name="time",
         Haplos=Haplos,desnames=desnames,designfunc=designftypes) 
names(out$coef) <- c(desnames,types)
out$coef
summary(out)