search

timothyfrasier / related

8 stars 7 forks source link

coancestry() resulting in fatal error with large datasets #16

Open fournier-c opened 1 year ago

fournier-c commented 1 year ago

Hi Timothy,

I have been using related successfully for the past two years with datasets consisting of 100-200 individuals and 10-15 loci. However, I am now trying to use it with a dataset consisting of ~500 individuals and 14 loci and when I use the coancestry function, it always returns a fatal error. I am using R 4.3.1.

Any help with this would be appreciated.

Thanks,

Cory

timothyfrasier commented 1 year ago

Hi Cory:

Yes, I'm afraid that this is a known problem that does not have an easy fix.

I wrote related quite a long time ago, when microsatellites were still the main marker of choice. As a result, I did not put any thought into how the package uses memory. This was never a problem until people started trying to use related to analyze SNP data sets (with hundreds to thousands of loci). Fixing this problem will require a complete rewrite of the package, which I have not had time to do. I am, however, hoping to get a student this academic year to try to help with this, so hopefully a fix is in the not-too-distant future (but still probably a few months).

-Tim

Timothy R. Frasier Coordinator: Forensic Sciences Program Professor: Biology Saint Mary's University 923 Robie Street Halifax, Nova Scotia B3H 3C3 Canada Tel: (902) 491-6382 E-mail: @.*** frasierlab.ca

From: fournier-c @.> Sent: Thursday, August 17, 2023 6:03 PM To: timothyfrasier/related @.> Cc: Subscribed @.***> Subject: [timothyfrasier/related] coancestry() resulting in fatal error with large datasets (Issue #16)

Hi Timothy,

I have been using related successfully for the past two years with datasets consisting of 100-200 individuals and 10-15 loci. However, I am now trying to use it with a dataset consisting of ~500 individuals and 14 loci and when I use the coancestry function, it always returns a fatal error. I am using R 4.3.1.

Any help with this would be appreciated.

Thanks,

Cory

— Reply to this email directly, view it on GitHubhttps://github.com/timothyfrasier/related/issues/16, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AA3NYWODBLDUVUWMKSM3J3LXV2BIPANCNFSM6AAAAAA3UUC4CA. You are receiving this because you are subscribed to this thread.Message ID: @.***>