I worked with someone (this is in astronomy) who said that papers in nature were the most likely to be wrong. They are in nature because they have a dramatic (new/unexpected) result. One good reason for a new/unexpected result is a mistake somewhere.

Edit: I'm not saying anything about this paper. I know nothing about this. Just a meta comment that, in really hard to get published in journals, there might be a bi-modality of papers. Really important and really wrong :)

It likely says something that, as an epidemiologist, the only way I've ever gotten a paper into Nature/Science is during a scary infectious disease epidemic. The rest of the time, they're largely uninterested.