QTL-QTL interaction

nata...@gmail.com

unread,

Jul 5, 2021, 5:48:59 AM7/5/21

to R/qtl discussion

Hi

I don't understand how can a/d be larger than 1 in the case of a binary trait. And how is the SE so big?

These are my fitqtl results after addqtl and addpair:

$FOB3_bin
fitqtl summary
Method: Haley-Knott regression
Model: binary phenotype
Number of observations : 118
Full model result
----------------------------------
Model formula: y ~ Q1 + Q2 + Q3

est SE t

Intercept -0.2722688 0.2461408 -1.10615077

L...@30.0a 0.5833693 0.3649499 1.59849135

L...@30.0d 2.2229040 0.6213122 3.57775695

LG...@6.6a -1.8501205 14.6417859 -0.12635894

LG...@6.6d 10.3528849 14.6465690 0.70684711

LG...@21.0a 0.5553481 14.6439920 0.03792327

LG...@21.0d -8.9403417 14.6415530 -0.61061431

thanks

Nataly

nata...@gmail.com

unread,

Jul 5, 2021, 5:57:25 AM7/5/21

to R/qtl discussion

basil=cross

s.aq=summary(addqtl)

here is the code I used for my 3 binary traits:

for(i in 1:3){

p<-phenames(basil)[i+n]

if(length(qtlist.bin[[p]][,1])>0){

q<-rbind(qtlist.bin[[p]][,-c(3,4)],s.aq[[p]])

rqtl2.bin[[p]]<-refineqtl(basil,p,makeqtl(basil,q[,1],q[,2],what="prob"),maxit.fitqtl=1e+6,tol=0.05,method="hk",model="binary")

out.ap.bin[[p]]<-addpair(basil,q[,1],p,rqtl2.bin[[p]],maxit=1e+6,tol=0.2,method="hk",model="binary")

qtlpairs[[p]]<-summary(out.ap.bin[[p]],thresholds=m.bin[p,])

s.fq[[p]]<-summary(out.fq[[p]]<- fitqtl(basil,p,rqtl2.bin[[p]],maxit=1e+6,tol=0.01,method="hk",model="binary",get.ests=T))

}

Is it because of the maxit / tol parameters?

ב-יום שני, 5 ביולי 2021 בשעה 12:48:59 UTC+3, nata...@gmail.com כתב/ה:

Karl Broman

unread,

Jul 5, 2021, 7:09:55 AM7/5/21

to R/qtl discussion

With the binary trait model, it is using logistic regression, so there are no constraints on the parameters.