Gavin Band writes:
I wondered what you (or your readers) make of this. Some points that might be of interest:
– The effect we discover is massive (OR > 10).
– The number of data points supporting that estimate is not *that* large (Figure 2).
– it can be thought of as a sort of collider effect – (human and parasite genotypes affecting disease status, which we ascertain on) – though I haven’t figured whether it’s really useful to think of it that way.
– It makes use of Stan! (Albeit only in a relatively minor way in Figure 2).All in all it’s a pretty striking signal and I wondered what a stats audience make of this – maybe it’s all convincing, or maybe there are things we’ve overlooked or could have done better? I’d certainly be interested in any thoughts…
The linked article is called “The protective effect of sickle cell haemoglobin against severe malaria depends on parasite genotype,” and I have nothing to say about it, as I’ve always found genetics to be very intimidating! But I’ll share with all of you.
Admittedly, I know nothing whatsoever about malaria or sickle cell haemoglobin. However, I note, assuming I counted correctly, that this paper had 25 authors. Is this typical in this area of research?
Paul:
I have no idea, but I do remember seeing the film Hemo the Magnificent in school several times when I was a kid. I just looked it up on wikipedia, and . . . it was directed by Frank Capra! And it featured the voice-acting legends Mel Blanc and June “Rocky” Foray! The whole thing’s on Youtube but I’m afraid to watch it and get disillusioned.
Wow! Thanks for the recommendation. “Hemo the Magnificent” was … well … magnificent!!!
I have no knowledge of the content of this article either. But regarding style, I don’t think that 25 authors is that unusual – but what about 59 references to Supplementary Methods or Supplementary Figures in the main text? How is anybody supposed to read an article like that? At least there are less than 59 decimal places in the p values (though still too many for my taste).
It might be interesting to model the improvement in survival from malaria for the heterozygotes versus the perniciousness of the homozygous state. I recollect that prior to recent advances, the incidence of the heterozygous state (AS hemoglobin) was 12-14% in Jamaica. The median survival was age 18 for homozygous sickle cell disease (SS hemoglobin) in Jamaica. Successful pregnancy outcomes were rare in SS women. Now of course, these numbers are medians with considerable range. I would expect a range of outcomes.
Jamaica got studied by two Brits, David Weatherall and J.B. Clegg.
I’m an evolutionary biologist, so I can comment on both the substance and style. Sickle hemoglobin is a well-known case of what’s known as “heterozygote advantage.” If you have one copy of sickle hemoglobin, S, and one copy of ‘regular’ hemoglobin, A, you do *not* get sickle cell anemia, and you’re pretty much immune to malaria. Thus the ‘fittest’ genotype for an individual to be is AS (the heterozygote); AA and SS suffer from malaria or anemia, respectively. This is shown quite convincingly in the paper: in the Gambia, about 15% of people are AS, but less than 1% of people who get severe malaria are AS. (This is not the point of the paper– we’ve long known this.) From a population genetic perspective, this is a case where natural selection acts to maintain genetic variability in a population (although this is only true where malaria is prevalent– in a non-malarial environment, AS individuals have no advantage).
The point of the paper is that there are genetic variations in the malarial parasite that seem to allow it to overcome (a little bit) the immunity of those with sickle hemoglobin. This makes total sense– parasites and pathogens often evolve ways of evading the defenses of their hosts. The statistical question that I believe the authors want to raise here at the blog is that despite having thousands of severe malaria patients at each of their study sites, there are so few cases among resistant AS and SS individuals– just 19 in the Gambia– spread across 9 different genetic types of parasitic infection, it’s hard to tell if the rates of infection are really higher in AS and SS individuals by one of the ‘mutant’ parasites as opposed to parasites that lack the mutant genotypes (see their Fig.2). This, I think, is the question posed to the blog– do they really have good evidence for this?
As far as style issues go, it is true that the number of authors on papers has been going up. This may be due to increased collaboration, or a thinning out of what one must do in order to deserve/demand/be offered co-authorship. I suspect all of these things play a part.
As far as the continuous references to “Supplementary” items in the paper, let me climb on my hobby horse. The paper has already been published in Nature (link at Biorxiv). It’s not quite 6 pages in the journal. If you get the pdf from Nature, there are an additional 5 pages of methods, 9 pages of “Extended Data” figures, and 3 pages of a “Reporting Summary”. If you have an actual copy of the journal, you don’t get any of these last 17 pages. (Andrew: you should definitely look at the “Reporting Summary”. It’s got 10 checkboxes the authors must check to insure the statistics have been done right!)
But the 17 pages of extra material in the pdf version is *not* the supplement! If you go back to Nature online, if you scroll way down you’ll find a link to a 32-page “Supplementary Information” with methods, figures (now “Supplementary” rather than “Extended”) and text, plus a separate link to “Supplementary Tables” in an Excel file with 9 table sheets in it.
It can be very hard to read carefully a paper that has its content spread over multiple files in multiple locations. The cheapness of online storage probably encourages authors to include more rather than less, but this allows dilution and obfuscation. And how stable are these links and storage? I once heard Edward Tufte ask, while holding aloft a 400-year old copy of a book by Galileo, “Will your website be up tomorrow?” If someone (or a large group of someones) has carried out a project worthy of a monograph, maybe they should publish a monograph.
It shows ~90% sickle cell children diagnosed with severe malaria were infected with at least one of three types of mutant parasites.That is compared to ~40% of non-sickle cell children.
They want to jump from that to a conclusion about evolutionary adaptation, which usn’t really warranted. The mutations may only increase the chances of severe disease in those with sickle cell. Ie, there may be a pool of children with mild symptoms who got excluded.