It's literally in the first sentence:
"The authors of a study that claimed to find a link between typical male brain anatomy and autism spectrum disorder (ASD) have retracted it and replaced it with a dramatically changed version after finding an error in their methodology."
> In the letter, Ecker wrote that she and her coauthors had "identified a problem with the script that was used for the prediction of biological sex in male and female individuals with ASD based on their respective neuroanatomy in our original sample."
> That meant that "the probabilities we assumed to be indicative of the default (male) category in the female individuals with ASD, as shown in Figure 1B of our article, were flipped and in fact reflect the probabilities for the category of biological female individuals."
Shitty academia code strikes again. It's sort of unfortunate that most of academic fields have turned to domain-specific software engineering, which graduates are woefully unprepared to do properly.