calculate multiple qtl with covariate and interacting covariate

[zahrap...@gmail.com]

Dear Dr. Broman 

I want to calculate multiple qtl analysis while considering covariate and interaction covariate .in the scanone analysis for example I used this script 

QTL_hk2<-scanone(QTL__root_length_TREATMENT_1, pheno.col=2,method="hk",addcovar = RL_C,intcovar= RL_C )

 and the results showed strong evidence of one qtl on chr=5

for multiple qtl analysis   I used below scripts: 

M_QTL<- calc.genoprob(QTL_root_length_TREATMENT, step=1)

M_qtl <- makeqtl(M_QTL, chr=c( "1","5"), pos=c(19, 53), what="prob")

summary(M_qtl)

out.fq <- fitqtl(QTL__root_length_TREATMENT, qtl=M_qtl,pheno.col=2, method="hk",formula=y~Q1+Q2)

summary(out.fq)

summary(fitqtl(QTL_root_length_TREATMENT, qtl=M_qtl, pheno.col=2, method="hk", get.ests=TRUE, dropone=FALSE))

covar <- data.frame(RL_C=QTL__root_length_TREATMENT$pheno$RL_C)

addcovarint(QTL_root_length_TREATMENT, pheno.col=2,M_qtl, covar,"RL_C" , formula=y~Q1+Q2, method="hk" )

But the result was weird, they didn't show any significant qtl . 

My question is : am I used  the correct way to add covariate and interacting covariate in multiple qtl analysis ? 

Is there any other function for calculate multiple qtl with covariate and interacting covariate ? 

Thank you in advance 

Zahra PAKBAZ

[Karl Broman]

In what way was the result weird? 

You can investigate interactive covariates in fitqtl() by comparing results with different formulas. That's what addcovarint() does: it adds a QTL x covariate interaction and compares the fit of the model with and without that interaction.

karl

--
You received this message because you are subscribed to the Google Groups "R/qtl discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rqtl-disc+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/rqtl-disc/b3b4b240-6b3b-4ebf-87e7-d3df2756177en%40googlegroups.com.

[zahra pakbaz]

It is weird for me because it shows the weak effect for  5@53.0:RL_C  while the Lod of qtl on chr 5 is around 7 when I add covariate and interaction covariate in scanone analysis. And I think it is a good lod score  and I can be sure that there is qtl on chr5 and also, there is covariate effect. (actually, RL_C is referring to root length data in the control condition ). Now I am confused by the results 

covar <- data.frame(RL_C=QTL_root_length_TREATMENT$pheno$RL_C)

> addcovarint(QTL_root_length_TREATMENT, pheno.col=2,M_qtl, covar,"RL_C" , formula=y~Q1+Q2,

+             method="hk" )
Method: Haley-Knott regression
Model:  normal phenotype
Model formula: y ~ Q1 + Q2 + RL_C

Add one QTL x covar interaction at a time table:
--------------------------------------------
                         df  Type III SS      LOD         %var       F value       Pvalue(Chi2)     Pvalue(F)  
1@19.0:RL_C   1       12.66      0.001671    0.002855    0.007492        0.930    0.9311  
5@53.0:RL_C  1     5256.90     0.699603    1.185105    3.163445        0.073    0.0769 .
---   
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

best regards 

Zahra Pakbaz 

Virus-free. www.avast.com

To view this discussion on the web visit https://groups.google.com/d/msgid/rqtl-disc/CAEwE18H99HfBkeFkmkOf0U5uLtcOM884A6njpy28Xb3wUq6TRw%40mail.gmail.com.

[Karl Broman]

The LOD scores from addcovarint() are for tests of the QTL x covariate interaction. In scanone, you'd take the difference between the LOD score with RL_C as interactive covariate and the LOD score when it is just an additive covariate. That the LOD score for the QTL x covariate interaction is small suggests that there is no evidence for the QTL effect being modified by RL_C value.

karl

[zahra pakbaz]

I really appreciate you. Your comment is perfect and always your answers helped me.
 May I ask you to suggest an article or something else, so I understand completely this subject? 

best regards 

Zahra Pakbaz

Virus-free. www.avast.com

To view this discussion on the web visit https://groups.google.com/d/msgid/rqtl-disc/f96a717d-6ccc-4eb5-a68e-8060d2122e45n%40googlegroups.com.

[Karl Broman]

Chapter 7 of the R/qtl book talks about covariates, including QTL x covariate interactions.

https://rqtl.org/book

karl