If you don't think mindkilling sets in until the general election, then you didn't actually follow the 2008 Democratic Primary (among other primaries, but it was on display in an extreme way in that particular case).
Also, in the specific case of sociopathy, I think that in response to a bad genetic test, the candidate would have an FMRI to measure the white matter in the amygdala, give an interview for the Hare Psychopathy test, and present any other confounding evidence that comes to hand, and then the debate would proceed much as MinibearRex suggests. Partisans would simply selectively weight the tests that supported their pre-existing intuitions.
I wrote an article for h+ predicting that the rapid fall in the cost of gene sequencing will allow U.S. voters to learn much about presidential candidates' DNA. The candidates won't be able to stop this because:
DNA analysis has a decent chance of reducing political bias by providing objective information about candidates. If, for example, 70% of the variation in human intelligence is determined by identified genes then DNA analysis would reduce disagreements among informed voters over a candidate's intelligence.