0% of model SNPs used Error in S-Predixcan

[Anmol Singh]

Hello,

I am trying to use the S-Predixcan software with the ARIC model to predict expression from my GWAS. I am getting this error that 0% of the model SNPs are being used, however when I input the model file into R I see that there is overlap between the model SNPs and my GWAS, ~ 35K snps, as per the merged_gwas_weights.png attachment. Do you have any suggestions for me to try as to why this error is occurring? Thanks, I have attached some screenshots of the files as well as the command I used and the error message.

[Hae Kyung Im]

Festus, do you have any recommendations ?

On Thu, Feb 8, 2024 at 2:01 PM Anmol Singh <singh....@gmail.com> wrote:

Hello,

I am trying to use the S-Predixcan software with the ARIC model to predict expression from my GWAS. I am getting this error that 0% of the model SNPs are being used, however when I input the model file into R I see that there is overlap between the model SNPs and my GWAS, ~ 35K snps, as per the merged_gwas_weights.png attachment. Do you have any suggestions for me to try as to why this error is occurring? Thanks, I have attached some screenshots of the files as well as the command I used and the error message.

--
You received this message because you are subscribed to the Google Groups "PrediXcan/MetaXcan" group.
To unsubscribe from this group and stop receiving emails from it, send an email to predixcanmetax...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/predixcanmetaxcan/b1ddff25-68e4-422b-a5c4-8e5f16a8dcd2n%40googlegroups.com.

[Festus]

You need to add this argument --keep_non_rsid in your command line since your variants are not rsids.

[Anmol Singh]

Great, that worked. Thank you!!