Re:Genes, blood type tied to risk of severe Covid-19
by bobcostas55 on Thursday June 18, 2020 @02:14PM
(#23565893)
>As far as GWAS studies go most turn out to be nonsense anyway.
[citation needed]
Re:Genes, blood type tied to risk of severe Covid-19
by glofish on Thursday June 18, 2020 @02:35PM
(#23566145)
A continuously updated list of bad genetics studies
https://massivesci.com/notes/gwas-complex-behavior-bad-biorx...
the only reason why not most GWASare listed there is because it does not pay to prove that a GWAS is a nonsense,
it is way too much work to disprove, and you don't get paid/supported for doing so
Re:Genes, blood type tied to risk of severe Covid-19
by drocer88 on Thursday June 18, 2020 @02:48PM
(#23566282)
Were any of those particular papers disproven, retracted or were not able to be reproduced?
This paper , "Genome-wide association studies: the good, the bad and the ugly" : https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4952840/ claims "the vast majority of associations identified by GWASs are extremely robust statistically and are reproducible in additional studies.".
Then overall health, how many other pre-existing conditions and ongoing health issues patients already have.
Then there is the dose and the exposure the patient has.
These are by far the largest contributors.
I find it hard to believe that they have successfully disentangled all these confounding factors.
Then and only conditionally perhaps the factors that the director lists. In all, I find the blog a staggering piece of how NOT to communicate, how not to send the wrong message. Most people will read this as: "blood type A is really bad"
GWAS in general is weak stuff, even in the cited paper the odds ratios are 1.5 and 1.77 which does not indicate a particularly strong effect. Statistically may be significant, but in a practical sense not actionable. As far as GWAS studies go most turn out to be nonsense anyway.