-
Notifications
You must be signed in to change notification settings - Fork 125
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Different results using GEMMA 0.98 vs. GEMMA 0.96 #188
Comments
@voichek please do share. |
@pjotrp I have sent you a mail a few days ago with the link (to [email protected]) |
@voichek it is not good practise to put E-mail addresses online. |
@voichek thanks for sharing. I agree there is something wrong with the results. I just reproduced the issue.
is the main difference. I'll need to git bisect until I get to a commit that gives the difference. I'll need to look into this and will make it a priority after 15/12. Can't really do it earlier because I'll be at http://2018.biohackathon.org/. Suggest you use 0.96 with this dataset for now. |
v0.97 gives the same result as v0.98. So it is between 0.96 and 0.97 something changed. |
I am sorry for that, I didn't know it was a problem |
I created a small test dataset which reproduces the problem with
The first line gives a unique name to each SNP. Now the head looks like
Now I can run git bisect to find the exact commit that creates the difference. |
@pjotrp I had an idea, can it be that it just calculate the associations for the wrong phenotype in the file? there are a few phenotypes and I am trying to take the first one... |
Certainly a possibility. It is easy to test by changing the number. i'll try tomorrow. |
So, commit 4c21290 which is official 0.96 renders NaN on my machine weather I build with atlas or openblas. Sadly 0.96 was created in a time we can not reproduce builds. Annoying. I am trying a more recent commit. |
So df1e049 is the commit that changed away from 0.96 results:
And this is the commit that fails f8d6ad4 |
Built and compiled with ~/.config/guix/current/bin/guix environment -C guix --ad-hoc gcc gdb gfortran:lib gsl eigen lapack atlas openblas zlib bash ld-wrapper perl ldc
make EIGEN_INCLUDE_PATH=$GUIX_ENVIRONMENT/include/eigen3 WITH_OPENBLAS=1 FORCE_DYNAMIC=1 -j 8 |
Hi,
I have run the same data with GEMMA version 0.96 and version 0.98. The results are different (0.98 are incorrect) for some reason.
The command line:
Results:
I am now uploading the data to my google drive and can share a link soon in a private mail.
Thanks for the help,
Yoav
The text was updated successfully, but these errors were encountered: