The problem is that different "races" are more closely related than people think. The article's title is misleading. Targeting one gene variation, say a cell surface receptor, will likely also have some cross-specificity to other variations although maybe in lower levels, that's all. Plus, microbes evolve after introduction into humans, allowing for what was originally specific for a gene variation found in, say people of relatively pure Nordic decent, to spread into strains that have multiple specificities. Nonethless, "gene targeting" would only be effective in homogenous populations like certain mongoloid and negroid groups -- Koreans, Japanese, Aborgines, etc. Groups that have either been geographically or politically isolated for long periods of time. Most Negroids and Caucasoids, because of higher migration rates all over the world throughout history, tend to be very heterogenous. Even the Chinese have significantly varying physical traits. This is why outbreeding between different ethnic backgrounds is good. You get variation that helps survive catastrophes. Even just different ethnic backgrounds within the same racial strata can get you a lot of variation. Like North/South/Eastern European, South/East Asian, East/West African, etc.
A more realistic and practical concern is the creation of highly virulent microbial strains. Like treating B. anthracis with high amounts of common antibiotics in someone's garage to select for strains that are resistant to immediate medical intervention.