It seems like everything is coming in twos the past couple of days. Yesterday we mentioned two books on the evolution of genomes and two stories involving either Wolbachia or sex determination. Today, we have two stories involving criticisms of scientific papers. One deals with the evolution of the bacterial flagellum, and the other addresses natural selection on the brain expressed gene ASPM in humans.
The first story involves everyone's favorite irreducibly complex cellular apparatus: the bacterial flagellum. During the Kitzmiller v. Dover trial, Nick Matzke worked with the plaintiffs (the pro-science side) to help them show intelligent design is not science and that the ID arguments are crap. Part of that involved debunking Michael Behe's argument that the bacterial flagellum could not have arisen via the conventional mechanisms of evolution, therefore requiring divine guidance. Afterwards, Matzke, along with Mark Pallen, turned the evolution of the flagellum story into a nice review paper.
Along comes Howard Ochman who, with the help of a post-doc, does a few blast searches and concludes that all genes encoding flagellar proteins share a common ancestor. While Pallen and Matzke pointed out some common ancestry of the genes encoding flagellar proteins, they did not claim that all genes shared a common ancestor which would have encoded some ur-flagellar structure. Matzke did some poking around into Liu and Ochman's methods and concluded they're bunk (and, upon further review, decided they're even more bunk than he originally thought). Long story short, Liu and Ochman used really crappy search parameters for identifying homologous proteins.
How or why did the Liu and Ochman paper get published? Upon first glance, one may conclude that it's the fault of PNAS's system of having National Academy members contribute papers without peer review. But Ochman is not a member of the NAS, and the paper was submitted Track II, meaning it was handled by an Academy member, but sent out for peer review like most scientific papers in most publications.
What went wrong? First off, the editor for the paper was Francisco Ayala, who, while an accomplished evolutionary geneticist, does not work in the field of genomics and probably does not spend much time thinking about identifying gene families. Ayala, unfamiliar with this field, may have sent the paper to unqualified reviewers who rubber stamped the publication without a critical review. In my opinion, the ideal editor would have been Masatoshi Nei, who specializes in the evolution of gene families. The criticism levied on the Ochman paper by Matzke and others is not an indictment on all models of the evolution of the bacterial flagellum, but a criticism of Liu and Ochman's particular model (and of their methods).
The second paper questions Bruce Lahn and colleagues' conclusion that the gene ASPM was under recent natural selection in humans (published here). Lahn's group used coalescent simulations to determine that the pattern of polymorphism at ASPM is inconsistent with any possible neutral models of molecular evolution. Therefore, natural selection has most likely acted on this gene in the recent past.
That conclusion was poorly challenged by Sally Otto and colleagues, and Lahn's group was able to brush it off without much sweat (Otto paper here, Lahn and colleagues' response here). But a new paper, from a group led by David Reich, challenges Lahn's result in a much better manner (reviewed here by p-ter at GNXP). Rather than using coalescent simulations to determine the neutral expectation, Reich's group used an empirical distribution generated from publicly available polymorphism data (ENCODE). They found no evidence for non-neutral patterns of polymorphism at ASPM, concluding that there is little evidence for recent natural selection at the locus.
Both of these debates illustrate the scientific process. The flagellum story shows that some crap does make it through peer review, but it gets caught by other people who criticize it using data. This debate is carried out in the scientific literature (and, now, on science blogs), where both parties present their most convincing evidence for their case. The ASPM debate shows the importance of reproducibility in science. A convincing result will be robust to different methods of inquiry. Lahn's conclusion that natural selection has acted on ASPM does not hold up when you search for selection using an alternative (yet equally valid) approach.
The flagellum story shows that some crap does make it through peer review, but it gets caught by other people who criticize it using data.
And goes further to show that "peer review" is not solely a pre-publication process. Peer review begins with the journal's procedures, but does not stop when a paper is published. For the most part it's what happens to a paper after publication that's most important.