search

ZuchnerLab / Maverick

A Mendelian approach to variant effect prediction built in keras
MIT License
16 stars 6 forks source link

Maverick seems to not filter out known benign variants #11

Open Qmphcm5l opened 1 week ago

Qmphcm5l commented 1 week ago

Hi, I want to use maverick to hunt for genes that cause auto inflammatory diseases. For this I got a bunch of whole genome variant vcfs of patients with those diseases. When I run my vcf files both with the maverick lite and full mode, my final scores contain a bunch of variants that are known benign variants on gnomad like CCDC66 3-56557250-T-TGGGGTAAGCA. This seems confusing to me as gnomad reference files are available for maverick. The input vcfs have been properly prepared for the use with annovar and the coordinates of the out put files line up with the coordinates of the input vcfs.

Qmphcm5l commented 1 week ago

Update, it seems like after having a look at the gnomad211_grch38_exons_af.txt that those variants are not present in that database. Is there any way to use a more updated gnomad dataset which includes those missing variants so that these variants are filtered out and i only get my unique mutations?