Here is a taste of the skepticism:
I have shared the paper with a couple of top notch human geneticists - who, in turn, thought that the paper was a bit of a joke. As one pointed out, if this was the profoundly important finding that NEJM's editorial claimed it to be, then publishing it in NEJM would be akin to Watson and Crick publishing their 1953 paper in Popular Mechanics rather than Nature!And another comment on the stats:
I believe at least one of these individuals is preparing a formal critique for publication. However, he has already flagged a number of problematic issues, including the Figure 1 data, the large number of unattached technical labels, and the failure to report whether those with the mutation suffered from other far more serious conditions (a likely result of lysosomal enzyme disease is developmental regression). As we know, Down's Syndrome patients have been reported to have an abnormally high incidence of stuttering, but their speech disorder is a relatively minor part of their condition. And, of course, not all persons who stutter have Down's Syndrome. You suggest that what Kang et al. might have found is a sub-type of the stuttering population (maybe,
like those with Down's Syndrome?), but that is NOT what they claimed in their paper. Anyhow, let's wait to see how this plays out when others with more expertise than me formerly comment on this paper.
By coincidence, a few days ago Nature carried the attached paper describing the unraveling of the Zebra Finch genome. They observed that more than 800 genes were functionally associated with the production of the Zebra Finch's learned song. And we're supposed to believe that one mutation in one chromosome site is functionally related to stuttering?
Regarding the 6%: The BRCA-1 mutation accounts for about 5% of breast cancer cases. We already know that stuttering is not a simple Mendelian trait with a single locus controlling expression. I would expect to find many loci, each representing independent paths to produce the same phenotype.
Regarding the Chi Square test: there are two problems in the figure. First, some subjects with the normal genotype are stutterers. Second, not all carrying the mutations are stutterers.The text asserts that the former individuals get their stuttering from another source. In the latter case, not all carriers need express the phenotype. From twin studies, we know that penetrance in stuttering is not 100% That is, not all individuals carrying the same genotype will express the same phenotype. This is going to confound the Chi Square test, which is a measure of how well phenotypes match genotypes. This will probably always be true in such studies of stuttering inheritance. Non-genetic effects - including fetal and juvenile brain development - may play a part in phenotypic expression. Social conditions may also play a part.
One could argue that such explanations are speculative excuse-making. They are speculative, but I think they are reasonable. The numbers are shaky, but that's the nature of inheritance outside of Biology 101 laboratory assignments. I still need to see some biology to make sense of the connection between the lysosomal proteins and speech, but that's another matter.
I am also a bit concerned. They did this Chi Square test which didn't turn out to be remarkable, but then they found a very good correspondence between a mutation in the lysosomal genes and stuttering (except for one female). I am not sure how this last observation can be included in a Chi Square test to prove that the statistical significance has increased to a discovery beyond doubt.
We need one or two other genetics team to have a true debate on this result. And we should always remember that the results significantly depend on the reliability of the lead on-hands scientist which I believe is Kang. If he has messed up, unwillingly or knowingly, we will be chasing a phantom.
5 comments:
I am not going to purport that I understand everythingin this post. But I do have a comment on the incidence of stuttering in persons with Down Syndrome.
Two+ years ago, I worked for an agency that supports people with developmental disabilities. (I now serve on thier Board of Directors).
The program I worked with supported about 50 clients, more than half having DS. Of thoe 28 or so people, 3 stuttered. As a person who stutters myself, I talked with them about it. None had had any services at all for the stuttering.Two were pretty severe with very obvious secondaries, like facial tension and grimaces. People just assumed that they weren't bothered by the stuttering, as it was the least of their problems.
Turns out, they were bothered by it. One felt her stuttering ws more of a challenge than her mild DS. I referred all three tot he local college that thas a fluency program, and all 3 got the benefit of individual and group speech sessions.And it was free.
The masters level students that worked with them also gothe chance to work with people who stutter who also have special needs.
I learned a lot from that experience, largely that people should never assume that just because a person hasd a developmental disability, that a speech disorder wouldn't have the same affect on them as it would a person who stutters who deosn't have an intellectual disability.
I like the new blog layout!!
It's quite wrong to say that NEJM is lower in prestige in relation to Nature. Its impact factor is higher than that of Nature (50.017 x 34.480).
If you research a little before repeating nonsense, you'll find that NEJM is the oldest continuously published medical journal in the world, and is the most widely read, cited, and influential general medical periodical in the world.
Dear Kang,
I was just cutting and pasting two emails that I got from someone with a very high academic rank and from a geneticist.
Would you rather know about these comments or not? You have been able to put forward your opinion publicly without having to pay a subscription! ;-)
In any case, you are bypassing the real debate. It is not on the quality of the journal, but the robustness of the results. Both NEJM and Nature have f*****d up before.
I don't care about the journal at all, but I very much care about the statistics and about reproducibility. And I am not 100% confident, but I am not a geneticist. And I definitely don't trust geneticists to be very good at stats especially if the disorder is not straight forward. Most geneticists are "lab animals".
So how about the statistical arguments?
Best wishes,
Tom
The reply by Kang makes the validity of the results even more suspicious. Instead of answering the concrete questions he defends his research by comparing impact factors. This sounds as naive as "I got my PhD from Harvard, a Nobel prize winner supervised me, how do you dare to question the validity of my results."
Leading journals are leading because they publish high quality papers, not vice versa.
Do not be stupid. You are completely wrong. I am not Changsoo Kang from NIDCD.
Post a Comment