problem with rtassel concerning a probable filtering of markers depending on the p-value

[Pierre]

Hi,

I am using rTASSEL v0.10.0 on an UNIX server.

I want to use it to analyze 7 traits measured on ~150 lines with ~3000000 markers showing MAF > 0.05.

When I try to do the GLM analysis, without using structure and kinship correction (I still did not try using these corrections) using the following instructions:

  GLMtest <- rTASSEL::assocModelFitter(

        tasObj = gftest,            # <- our prior TASSEL object

      . ~ .,                     

      fitMarkers = TRUE,             

      kinship = NULL,

      fastAssociation = FALSE

    )

the result shows a huge reduction in the number of markers

  ..@ results  :List of 2

  .. ..$ GLM_Stats    :'data.frame': 443956 obs. of  18 variables:

  glm_df <- GLMtest@results$GLM_Stats

  nrow(glm_df)

  [1] 443956

and this set of markers contains only the markers showing P-values below 0.001 for at least one of the traits:

summary(glm_df$p)

      Min.        1st Qu.       Median      Mean        3rd Qu.        Max. 

0.0000000 0.0001853 0.0004083 0.0004383 0.0006808 0.0010000 

As a results, and for each of the 7 traits, the QQ-plot and Manhattan plot show only markers having -log(P-value) > 3.

Why is it like that ?

What can be changed to have all the markers appearing in the results, including those that have a P-value comprised in [0.001 - 1].

Thanks in advance for your feedback.

Pierre

[Brandon Monier]

Hi Pierre,

When running the assocModelFitter method, have you manually changed the default maxP parameter? It defaults to "0.001". This will report only markers with P-value <= to the given value. More info can be found here:

https://rtassel.maizegenetics.net/reference/assocModelFitter

Best,

Brandon M.

[Pierre Berthomieu]

Dear Brandon,

Thank you for your feedback. I did not change the maxP parameter. I had not found that information anywhere. 

Now that I changed it, it is working fine.

Thanks a lot.

Best regars

Pierre 

--
You received this message because you are subscribed to a topic in the Google Groups "TASSEL - Trait Analysis by Association, Evolution and Linkage" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/tassel/yKDM48hbBME/unsubscribe.
To unsubscribe from this group and all its topics, send an email to tassel+un...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/tassel/e51dfbdd-3293-4ba4-9843-204b6c7d5351n%40googlegroups.com.