How to compile a list of non-cancerous mutations

Question

[Gene_MMP8]

I am trying to perform a one-class classification of driver vs non-driver mutations. I have a list of experimentally validated driver mutations from a pancancer study. But is it possible to retrieve a list of non-cancerous mutations as well? Since the number of non-cancerous mutations far exceeds the number drivers, I would train a one-class SVM on the list of non-cancer mutations and predict the outliers as drivers. But before that, I need a list of such high confidence non-cancer mutations. Is there a way to compile a list from different databases?