Context specific eQTL with no overlapping individuals
Closed this issue · 4 comments
mariesaitou commented
Hello. Thank you for developing this great tool.
I was able to run this for the toy data, but I realized that I could not use this tool for my dataset as is because my data are animals raised in three conditions, and I do not have data for multiple conditions of the same individual.
They are genetically homogeneous populations that are divided into three separate groups. Is there any way to use this tool for such a data set? Or, do you know of a good tool for context-specific eQTL for non-overlapping individuals?
So far, I have tried PICALO but it did not work well in my environment.
I appreciate your help.
BrunildaBalliu commented
Hi Marie,
So glad you tried FastGxC!
Indeed, FastGxC is intended for cases where samples overlap across contexts
as it models the shared noise.
What is the exact test that you are interested in? Do you want to find
eQTLs with effect size heterogeneity in the three conditions? The extension
of FastGxC to the case of non-overlapping samples should be fairly trivial
(similar to an interaction test). Otherwise you can fit a genotype -
condition interaction model
Expression ~ a + b1 Genotype + b2 Condition + g1 Genotype x Condition
and test for the significance of g1 using a Wald test or LRT
Best,
Bruna
…On Tue, Nov 8, 2022 at 4:50 AM mariesaitou ***@***.***> wrote:
Hello. Thank you for developing this great tool.
I was able to run this for the toy data, but I realized that I could not
use this tool for my dataset as is because my data are animals raised in
three conditions, and I do not have data for multiple conditions of the
same individual.
They are genetically homogeneous populations that are divided into three
separate groups. Is there any way to use this tool for such a data set? Or,
do you know of a good tool for context-specific eQTL for non-overlapping
individuals?
So far, I have tried PICALO <https://github.com/molgenis/PICALO> but it
did not work well in my environment.
I appreciate your help.
—
Reply to this email directly, view it on GitHub
<#2>, or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABHAMQPYD2XHMV7GQZ5KLIDWHJEC5ANCNFSM6AAAAAAR2JBPTI>
.
You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>
mariesaitou commented
BrunildaBalliu commented
In this case, I would run a genotype-by-condition interaction model
Expression ~ a + b1 Genotype + b2 Condition + g1 Genotype x Condition and
test for the significance of g1 using a likelihood ratio test. If g1
significantly differs from zero, the genetic effect differs between at
least one pair of your conditions. Does that make sense?
The other option is to run a meta-analysis test.
…On Tue, Nov 8, 2022 at 1:14 PM mariesaitou ***@***.***> wrote:
Thank you very much for your prompt reply...!
Yes, "find eQTLs with effect size heterogeneity in the three conditions"
are what I am interested in.
Such as below - this is just one example.
[image: Screenshot 2022-11-08 at 22 06 32]
<https://user-images.githubusercontent.com/41928205/200675609-c7359a56-d73a-455a-8d8e-e3739dd10a60.png>
—
Reply to this email directly, view it on GitHub
<#2 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABHAMQJQ6Z7U54YXXX5YBJ3WHK7CPANCNFSM6AAAAAAR2JBPTI>
.
You are receiving this because you commented.Message ID:
***@***.***>
mariesaitou commented
Yes, I managed to make the model work on R for testing! Thanks a lot.