revqtl and fitqtl

[Eiram Elahi]

Hi Karl,

 

I am conducting MQM using data (BX 142 Mice, 112 markers chr1-19)

 

5 main effects QTLs were detected from scanone, and also came up as 2 epistatic pairs following twoscan.  I have now tried to fit these in the MQM model and I'm not sure how to describe the drop one analysis (I have ordered your book) and looked through tutorials but still a little lost.  I have basic statistical knowledge and know what F/P/Chi are; but I am unable to put this output into some meaning.  Please advise.

 

b) secondly, after revqtl function ; Q4 (@20cM) is significantly moved to 47cM only 3cM away from Q5 (16@50.1) can you help me understand what may be the underlying cause of this.  

 

I have pasted the output from the analysis below.

 

Kind regards,

 

Eiram

 

<dropone output>

 

                fitqtl summary

Method: multiple imputation
Model:  normal phenotype
Number of observations : 142

Full model result
---------------------------------- 
Model formula: y ~ Q1 + Q2 + Q3 + Q4 + Q5 + Q1:Q3 + Q2:Q5

       df       SS        MS      LOD     %var Pvalue(Chi2)    Pvalue(F)
Model   7 13250.95 1892.9928 13.65484 35.77881 4.000522e-11 1.255694e-10
Error 134 23784.80  177.4985                                           
Total 141 37035.75                                                     

Drop one QTL at a time ANOVA table:
---------------------------------- 
               df Type III SS     LOD   %var F value Pvalue(Chi2) Pvalue(F)  
2@60.0          2     2491.81 3.07215 6.7281  7.0192        0.001   0.00126 **
5@29.4          2     1409.45 1.77514 3.8056  3.9703        0.017   0.02113 *
5@64.4          2     1805.43 2.25600 4.8748  5.0858        0.006   0.00743 **
16@20.1         1     1021.93 1.29717 2.7593  5.7574        0.015   0.01780 *
16@50.1         2      219.85 0.28371 0.5936  0.6193        0.520   0.53985  
2@60.0:5@64.4   1      771.09 0.98379 2.0820  4.3442        0.033   0.03903 *
5@29.4:16@50.1  1       67.13 0.08691 0.1813  0.3782        0.527   0.53961  
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

qt to be fitted

      name chr   pos n.gen
Q1  2@60.0   2 59.97     2
Q2  5@29.4   5 29.39     2
Q3  5@64.4   5 64.39     2
Q4 16@20.1  16 20.07     2
Q5 16@50.1  16 50.07     2

 

and the revqtl output

 

      name chr   pos n.gen
Q1  2@55.9   2 55.88     2
Q2  5@34.4   5 34.39     2
Q3  5@64.4   5 64.39     2
Q4 16@46.8  16 46.83     2
Q5 16@50.1  16 50.07     2

 

Total LOD increase of 1.67

 

[Karl Broman]

On Nov 9, 2012, at 3:38 AM, Eiram Elahi <eiram...@gmail.com> wrote:

Hi Karl,

 

I am conducting MQM using data (BX 142 Mice, 112 markers chr1-19)

 

5 main effects QTLs were detected from scanone, and also came up as 2 epistatic pairs following twoscan.  I have now tried to fit these in the MQM model and I'm not sure how to describe the drop one analysis (I have ordered your book) and looked through tutorials but still a little lost.  I have basic statistical knowledge and know what F/P/Chi are; but I am unable to put this output into some meaning.  Please advise.

Here's the description in the tutorial on multiple QTL analysis (http://www.rqtl.org/tutorials/new_multiqtl.pdf):

"The initial table indicates the overall fit of the model; the LOD score of 21.8 is relative to the null model (with no QTL). In the second table, each locus is dropped from the model, one at a time, and a comparison is made between the full model and the model with the term omitted. If a QTL is dropped, any interactions it is involved in are also dropped, and so the loci on chr 6 and 15 are associated with 2 degrees of freedom, as the 6×15 interaction is dropped when either of these QTL is dropped. The results indicate strong evidence for all of these loci as well as the interaction." 

b) secondly, after revqtl function ; Q4 (@20cM) is significantly moved to 47cM only 3cM away from Q5 (16@50.1) can you help me understand what may be the underlying cause of this.  

I would look at plots of phenotypes vs the two-locus genotypes for these tightly linked QTL, using effectplot() and/or plotPXG().

karl

[Eiram Elahi]

Dear Karl,

 

Thank you for your help.  I understand the dropone QTL now.

 

Regarding the second issue, I am still unclear.

 

Following Scanone there were two peaks on Chr 16 @20 and 50cM and identified as two QTLs.  Though, I am not sure how to check if these are linked (in coupling) I used the 'allpairs=FALSE' and 'subset';

The results of which;

 

summary(subset(out2, chr=16))

        pos1f pos2f lod.full lod.fv1 lod.int     pos1a pos2a lod.add lod.av1
c16:c16  50.1  55.1     8.04    3.89    2.74      47.6  57.6     5.3    1.15

 

This shows strong evidence for 2nd QTL @55, but nothing @ 20  and so I concluded 20 and 50 as separate QTLs and not included the one @55 in further analysis were separate QTLs

 

I looked at effect plots for both sets, chr16 @ 50.1 and 55.1 and @20 and 50, please see word document for screen shots - my interpretation is that these are not linked on the basis that the effectplot lines do not intercept.

 

So I am not sure why in revqtl after the first iteration of QTL @20 was re-positioned @47, yet the second QTL 50 stayed @50.

 

Would appreciate any help in understanding this, as the QTL @20 is a gene specific to the parent which was used in the backcross so it is quite important.  

 

Kind regards,

Eiram

 

 

--
You received this message because you are subscribed to the Google Groups "R/qtl discussion" group.
To post to this group, send email to rqtl...@googlegroups.com.
To unsubscribe from this group, send email to rqtl-disc+...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/rqtl-disc?hl=en.

--
Eiram Elahi
Mobile: +65 8383 9636 | Email: eiram...@gmail.com

[Zeb]

Hi Karl,

I have a somewhat related question in which I think I know the answer but looking for confirmation. Say for example using scanone you identify 2 QTLs (for simplicity we will say they are located on separate chromosomes). You decide to assign an interactive covariate to the marker that lies beneath the peak of QTL #2 and find significant interaction between this covariate and QTL #1. Now you decide to run scantwo (with no covariates). The results of this analysis should indicate a QTL1 and QTL2 interaction similar to that found when using the interactive covariate in a scanone analysis, correct?

Zeb

On Friday, November 9, 2012 4:46:31 AM UTC-8, Karl Broman wrote:

On Nov 9, 2012, at 3:38 AM, Eiram Elahi <eiram...@gmail.com> wrote:

Hi Karl,

 

I am conducting MQM using data (BX 142 Mice, 112 markers chr1-19)

 

5 main effects QTLs were detected from scanone, and also came up as 2 epistatic pairs following twoscan.  I have now tried to fit these in the MQM model and I'm not sure how to describe the drop one analysis (I have ordered your book) and looked through tutorials but still a little lost.  I have basic statistical knowledge and know what F/P/Chi are; but I am unable to put this output into some meaning.  Please advise.

Here's the description in the tutorial on multiple QTL analysis (http://www.rqtl.org/tutorials/new_multiqtl.pdf):

"The initial table indicates the overall fit of the model; the LOD score of 21.8 is relative to the null model (with no QTL). In the second table, each locus is dropped from the model, one at a time, and a comparison is made between the full model and the model with the term omitted. If a QTL is dropped, any interactions it is involved in are also dropped, and so the loci on chr 6 and 15 are associated with 2 degrees of freedom, as the 6×15 interaction is dropped when either of these QTL is dropped. The results indicate strong evidence for all of these loci as well as the interaction." 

b) secondly, after revqtl function ; Q4 (@20cM) is significantly moved to 47cM only 3cM away from Q5 (1...@50.1) can you help me understand what may be the underlying cause of this.  

[Karl Broman]

Yes those two approaches will give you basically the same results. The only difference concerns the treatment of missing genotypes.

karl

> --
> You received this message because you are subscribed to the Google Groups "R/qtl discussion" group.

> To view this discussion on the web visit https://groups.google.com/d/msg/rqtl-disc/-/Jusb0rLP6G0J.