Let me start by suggesting that you're begging the question (I mean that in the traditional way, not in its contemporary misuse). The question, in this case: "Is there a predictive, genetically defined thing called race," and your answer is "yes."
---My Example---
Instead of being worried about interracial breeding, you could be worried about breeding between people of different hair color. People with light hair carry certain disease factors with them more often than people with dark hair (a true statement, and so is its converse). So, what effect will the interbreeding of people of different hair colors have?
Now, the purpose of this example is not to say that race is a useless factor; certainly, African American populations wouldn't be sought out in drug trials if it were. However, WHY it's useful is important to keep in mind. Race is useful because, when well defined, it can point to distinct geographic origins (isolated populations).
In my hair color example, it can be shown that hair color also (at one point before different groups mixed) would have some of the same associative and predictive qualities that race currently has. But we have genomic information now, so we don't particularly need hair color as an indicator of anything. Nor, I argue, do we need to use race as such.