What degree of influence do SNPs have on activity of ligands at receptors?

I know that generally, evolution tends to evolve towards having some wiggle room in respect to effect of polymorphisms on binding of endogenous ligands, but with synthetic ligands, especially modern structurally-unrelated molecules, does the same continue to hold true?

I’ve been particularly wondering about this in respect to biased agonists with selective functionality for some pathways activated by a receptor over others – wouldn’t activity of that degree of selectivity be particularly vulnerable to disruptions by SNPs?

Hypothetically, would a ligand preferentially activating therapeutic pathway A but not a highly hazardous pathway B of an ABC receptor be a viable drug at all? Or would it be prime "expect it to kill somebody sooner or later" material?