Pick one: Genomic responses in mouse models POORLY/GREATLY mimic human inflammatory diseases
About a year ago Seok et al. shocked the biomedical world with the verdict that mice are not humans, or more specifically, that the blood genomic responses in various inflammatory conditions do not correlate at all between human patients, and the corresponding disease models (see previous post , as well as this one). Now another paper, by Takao et al. and also published in PNAS, concludes the exact opposite, that is that there is a near perfect correlation between blood genomic responses in mouse and man.
Meanwhile, the initial publication is among the top cited medical publications of the last year, and hundreds of newspapers and blogs (including this one) have covered it. It will be interesting to see how much media coverage the Takao paper will receive, probably much less. But what happened, which paper should we believe?
It is interesting to note that Takao et al. reevaluated the same gene expression datasets used in the Seok study, but focused on genes whose expression levels were significantly changed in both humans and mice. In essence, the conclusion is that Seok et al. had introduced severe bias by including genes that showed a significant response only in the human conditions and not in mouse models. Their final verdict about the Seok paper uses language that it rarely seen today in scientific journals (but has a longstanding academic tradition!):
‘Taken together, it is not surprising that Seok et al. found almost no correlation between the genomic responses in human disease conditions and those of mouse models, because they used methodologies with limited detection ability, which are rarely used by researchers who study animal models. With reasonably sophisticated and commonly used methods, we found highly significant correlations between human and mouse data. It should be noted that in our analyses we used the data from the exact same studies as those used in Seok et al., and still we found highly significant similarities, demonstrating that their failure to detect correlations is purely a result of the inappropriately biased methodologies they used’.