Percentage of variation accounted for

[TimL]

Hi all:

I am working with a couple of quantitative geneticists who are seeing
output from BimBam for the first time (they have both used regression-
based methods extensively in the past). One question that I have not
been able to answer is whether Bimbam gives us some measure of the
percentage of variability (of the phenotype) that each SNP
explains.....perhaps this is "mu" from the mean of coefficients
output?

Any help would be appreciated,

Cheers!

Tim

PS: Have any of you used Bimbam with mouse data before or are we the
first?

[Yongtao Guan]

Tim,

That's a good question.
For each SNP, after running single SNP analysis, you obtain mu
(intercept), a (addition effect size) , and d (dominant effect size),
then you can compute predicted phenotype p_i for each individual i.
Hence the variance across individuals, var(p_i), can be computed, so
does var(y_i), the variance of observed phenotypes. Now the percentage
of variability of phenotypes that a single SNP explains is just the
ratio of var(p_i) / var(y_i).
However, one can not simply add up these ratios from different SNPs to
obtain percentage of variance explained by these multiple SNPs for
many reasons. An obvious one would be these different SNPs might be
correlated. Matthew and I just finished a paper whose main aim (or one
of the main aims) is to estimate percentage of variance explained by
all SNPs in a genome wide setting, stay tuned!

Grant

> --
>
> You received this message because you are subscribed to the Google Groups "BIMBAM HELP" group.
> To post to this group, send email to bimba...@googlegroups.com.
> To unsubscribe from this group, send email to bimbam_help...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/bimbam_help?hl=en.
>
>
>



--
Yongtao Guan, PhD
http://home.uchicago.edu/~ytguan/