search

weizhouUMICH / SAIGE

GNU Lesser General Public License v3.0
189 stars 73 forks source link

Error message: "ERROR! The first variance component parameter estimate is 0" #411

Open dkimsjch opened 2 years ago

dkimsjch commented 2 years ago

Dear Wei,

I received the following error messages when I ran the single-variant analysis for a quantitative trait.

Error in glmmkin.ai_PCG_Rcpp_Quantitative(bedFile, bimFile, famFile, Xorig, : ERROR! The first variance component parameter estimate is 0 Calls: fitNULLGLMM -> system.time -> glmmkin.ai_PCG_Rcpp_Quantitative

I tried both invNormalize=TRUE and invNormalize=FALSE and both including and not including covariates, but I had the same error message. For some different outcome traits with the same dataset, it ran well without this error message. I am using version 1.1.0.

Could you let me know what I can do to solve this issue?

Thank you! Dan

DurandAxelle commented 1 year ago

Dear Wei, I receive also the same error messafes when I ran the single-variant analysis for a quantitative trait with SAIGE version saige-1.1.6.2. Error in glmmkin.ai_PCG_Rcpp_Quantitative(bedFile, bimFile, famFile, Xorig, : ERROR! The first variance component parameter estimate is 0 Calls : fitNULLGLMM -> system.time -> glmmkin.ai_PCG_Rcpp_Quantitative I tried also invNormalize=TRUE and invNormalize=FALSE, but I had the same error message.

Could you let me know what I can do to solve this issue? Thank you! Axelle

surakshavinod commented 1 year ago

Hi, I'm also running into the same issue. Were you able to solve it?

Thanks, Suraksha

akhilpampana commented 1 year ago

Is this issue resolved?

AmelieScha commented 1 year ago

I'm running into the same issue. Did anyone figure out how to fix this? Any help is appreciated.

Thanks, Amelie

surakshavinod commented 1 year ago

Hi, unfortunately, I haven't been able to resolve this issue till date. Please let me know if an alternative has been found or if this issue has been resolved.

Thanks, Suraksha

vincent-mauduit commented 1 year ago

Hi,

I ran into a similar error because I was using the parameter --LOCO=TRUE when working with a sparse GRM. This parameter can be used exclusively when a full GRM was fit at the previous step. By default it is set to TRUE so you may need to add --LOCO=FALSE to your command line.

Hopefully it helps, Vincent