My fuckbrain PI forced me to publish a paper that was very clearly overinterpreted horseshit. It went against better data in our field, but fuck it, she was in it for the novelty. That shit went into PNAS, and I have a Science paper that's also obvious bullshit. Aside from those, I cannot tell you how many times I've used a paper's mathematical model and gotten results conflicting with those that were published. Mathematical model.