The way I see it, here we are not looking for associations between genes and a single-feature trait related to either face of fingerprints. If my guess is correct as to the nature of this question, the OP is looking to connect fingerprints to more than just a facial shape or eye color. Cool idea to be sure, but is it practical?
Let's assume that 5-10 features identify fingerprints at least somewhat uniquely, and that the same is true for facial features. I think these are conservative assumptions. Is it likely that there are statistically significant connections between each feature combination and genetic variations? I think not, despite this being a theoretically sound approach.