combining F2 and F3 generation
188 views
Skip to first unread message
Masato
Apr 10, 2012, 2:39:53 PM4/10/12
to R/qtl discussion
Dear Karl and Group members,
In my F2 generation (384 individuals), the only a couple of
individuals showed the high phenotypic level as of parent A, which, I
guess, could failed to provide enough power for QTL detection (Does
this assumption seem OK?).
I thus selected F2 individuals by their genotypes, which at least
have parent A homozygote at the closest markers to potential QTL
locations, and then let them cross each other (3 families in total) to
raise F3 generation (96 individuals). Luckily I could have ten F3
individuals with the almost same phenotypic levels as of the parent A.
My questions are:
(1) To make a linkage map, I should only use the F2 generation and
exclude F3 generation since the recombination ratio in F3 differs from
F2s, is this correct?
(2) to detect QTLs, can I combine F2 and F3 phenotype and genotype
data and run scanone, scantwo, and rather stepwiseqtl using above
linkage map? Don't I have to make any change in my genotype coding nor
else?
Thank you very much for your help always!
Best,
Masato
In my F2 generation (384 individuals), the only a couple of
individuals showed the high phenotypic level as of parent A, which, I
guess, could failed to provide enough power for QTL detection (Does
this assumption seem OK?).
I thus selected F2 individuals by their genotypes, which at least
have parent A homozygote at the closest markers to potential QTL
locations, and then let them cross each other (3 families in total) to
raise F3 generation (96 individuals). Luckily I could have ten F3
individuals with the almost same phenotypic levels as of the parent A.
My questions are:
(1) To make a linkage map, I should only use the F2 generation and
exclude F3 generation since the recombination ratio in F3 differs from
F2s, is this correct?
(2) to detect QTLs, can I combine F2 and F3 phenotype and genotype
data and run scanone, scantwo, and rather stepwiseqtl using above
linkage map? Don't I have to make any change in my genotype coding nor
else?
Thank you very much for your help always!
Best,
Masato
Karl Broman
Apr 12, 2012, 11:13:25 AM4/12/12
to rqtl...@googlegroups.com
F3 individuals won't give you increased power to detect QTL (vs F2 individuals). With the same number of individuals, the mapping precision would be better, but power would actually go down as the significance threshold would have to go up (due to the "expanded" genome).
One could potentially use the F3 individuals to estimate the genetic map and to detect QTL, but R/qtl doesn't include code to handle them. I would be most concerned about how to take account of the relationships among individuals.
You might consider using QTLRel (see http://www.biomedcentral.com/1471-2156/12/66).
karl
> --
> 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.
>
Masato
Apr 12, 2012, 4:04:09 PM4/12/12
to R/qtl discussion
Dear Karl,
Thank you very much for your thoughts and suggestion.
I'll then try out QTLRel with F2 and F3, and decide whether I should
drop F3 dataset or not.
I appreciate your clear and intuitive explanation too!
Thank you.
Best,
Masato
On Apr 12, 11:13 am, Karl Broman <kbro...@gmail.com> wrote:
> F3 individuals won't give you increased power to detect QTL (vs F2 individuals). With the same number of individuals, the mapping precision would be better, but power would actually go down as the significance threshold would have to go up (due to the "expanded" genome).
>
> One could potentially use the F3 individuals to estimate the genetic map and to detect QTL, but R/qtl doesn't include code to handle them. I would be most concerned about how to take account of the relationships among individuals.
>
> You might consider using QTLRel (seehttp://www.biomedcentral.com/1471-2156/12/66).
Thank you very much for your thoughts and suggestion.
I'll then try out QTLRel with F2 and F3, and decide whether I should
drop F3 dataset or not.
I appreciate your clear and intuitive explanation too!
Thank you.
Best,
Masato
On Apr 12, 11:13 am, Karl Broman <kbro...@gmail.com> wrote:
> F3 individuals won't give you increased power to detect QTL (vs F2 individuals). With the same number of individuals, the mapping precision would be better, but power would actually go down as the significance threshold would have to go up (due to the "expanded" genome).
>
> One could potentially use the F3 individuals to estimate the genetic map and to detect QTL, but R/qtl doesn't include code to handle them. I would be most concerned about how to take account of the relationships among individuals.
>
Reply all
Reply to author
Forward
0 new messages