--
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/8d4b0bba-90fe-4603-9db0-aec2009998c2n%40googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/predixcanmetaxcan/37877dc2-9c89-46e1-9dab-4e87d75c23c2n%40googlegroups.com.
Hi everyone,
This was really helpful. With the perspective of running MASHR models on UKB .bgens, we sought to test the manual mapping via --variant_mapping against --bgen_use_rsid in the EN models first.
We subsetted the c1 .bgen based on rsids using bgenix first (now containing the first 10 variants with unique rsIDs that were considered by the EN model).
Using --bgen_use_rsid worked flawlessly.
Next, we generated a tab separated file like so (alternate_ids obtained from bgenix -list as ID):
ID rsid
1:10000_A_G rs123
1:20000_C_G rs234
rs345 rs345
Finally, we used the command “--variant_mapping file.txt ID rsid”.
We obtain predictions considering only such variants which are stored as rsid (in this example e. g. rs345), but not for such which are encoded by chromosome:position_ref_alt.
We tested multiple things without success, including:
Did we potentially get the variant identifiers wrong (i. e. are they not what’s listed by the bgenix -list output under “alternate_ids”)?
Any help would be highly appreciated.
BW,
Hyunchan