[R] Fwd: CoxME: Family relatedness

Marie Dogherty marie.dogherty79 at gmail.com
Mon Sep 15 12:27:26 CEST 2014


Hello all,



I have a table like this, with ~300 individuals:



Famid     Id   Faid Moid Cohort    Sex  Survival Event SNP1 SNP2 SNP3

1    1    0    0    0    0    10   1    0    1    0

2    2    0    0    1    1    20   1    0    0    0

2    3    0    0    0    0    25   1    0    1    0

4    5    1    2    0    0    35   1    0    1    1

4    6    1    2    0    0    35   0    1    0    1







famid=family id, id=id of person,faid=father id,moid=mother id.



My question of interest: What impact does SNP1/SNP2/SNP3 have on survival
of individuals (Id), after accounting for possible effects due to family
relatedness (Famid).



So I want to account for family-specific frailty effects, and individual
frailty effects according to degree of relationship between individuals.



The commands I've used are:



Library(survival)

Library(coxme)

Library(kinship2)

Library(bdsmatrix)



Death.dat <-read.table(“Table”,header=T)



deathdat.kmat
<-makekinship(famid=death.dat$famid,id=death.dat$id,father=death.dat$faid,mother=death.dat$moid)



death.dat1<-subset(death.dat,!is.na(Survival))



all <-dimnames(deathdat.kmat)[[1]]



temp <-which(!is.na(death.dat$Survival[match(all,death.dat$id)]))



deathdat1.kmat <-deathdat.kmat[temp,temp]



model4
<-coxme(Surv(Survival,Event)~Sex+strata(Cohort)+SNP1+SNP2+SNP3,data=death.dat1,id|famid,varlist=list(deathdat1.kmat,famblockf.mat),pdcheck=FALSE)







I almost completely edited these commands from :
http://www.ncbi.nlm.nih.gov/pubmed/21786277 as I am new to R.



The error I obtain is:



Error in coxme(Surv(Survival, Event) ~ Sex + strata(Cohort) + SNP1 + SNP2
+  :

No observations remain in the data set

In addition: Warning message:

In Ops.factor(id, famid) : | not meaningful for factors





I have two questions:

1. What is the difference between (id|famid) and (1+id|famid)/How to I tell
which is appropriate for my data set/Have I formatted that section of the
command properly?

2. Does anyone understand the error/how to fix it?



Many thanks

	[[alternative HTML version deleted]]



More information about the R-help mailing list