I know these specialist-aimed paper can be hard to read and understand especially for people not used to looking at such things, but the basic conclusion is that all of the risk alleles combined only account for ~1% of BMI/weight variance.
All SNPs combined explained 0.9% of BMI variation, with an AUC of 0.574 (95% CI: 0.559, 0.590) for prediction of obesity.
The rest is due to other factors.
Like eating too fucking much. (That part is not in the paper.)
So nope, your genetics ain’t makin’ you fat.
That paper does not say that your genetics aren’t making you fat. It says that obesity has a very high heritability and that the 12 SNPs investigated in the paper in that population with a particular history explain 0.9% of the variance.
Obesity has a very high heritability. It is very, very, very, very probable that the genes you have are really important when it comes to your obesity risk. “All SNPs” in this paper is 12. That is not all of the SNPs associated with obesity by a long shot. Every time there’s a new genome-wide association study they find more alleles associated with obesity. By 2014 they’d found > almost 4x that, and with the massive improvements in sequencing technology, I imagine the list will get longer. Explanatory power is still low – in the 3-5% range. Obesity is a complex trait. There could be hundreds of alleles associated with obesity. So these 12 alleles explain a small amount of variation – that tells us about these SNPs and a little about the distribution of effect sizes of common obesity SNPs.
Newer estimates using more SNPs found with different association methods find they can explain ~20%. [General methodological weaknesses discussed by this SNP hater.]
The SNPs we find first are the very common ones with big effects, but most of the missing heritability is in the many, many loci of tiny effect.
Finding those tiny ones in human populations where you can’t do controlled experiments with lots of replication? I can’t imagine where you’d begin. It’s also very difficult to estimate actual effect sizes without controlled experiments. Twin studies really don’t cut it – when you’re looking at small effect sizes for many loci, you need enormous samples.
I emphasized common above because that’s something that can be easily overlooked in the context of these papers. Common SNPs are ones found across populations in very high proportions of obese people. But rare SNPs can be important too – imagine 10 obesity boosting SNPs at different loci, and assign 1 each randomly to half the population. Those SNPs get hard to find if they have small effects and are often tossed out of studies like this even if they are large because they aren’t “common.” (Though this example may be above the common threshold – I don’t know what it is typical here.)
So #1 big problem here – 12 SNPs are a tiny fraction of the SNPs associated with obesity. #2 This doesn’t even begin to touch on epigenetics.#3 It only considers additive genetic effects, not epistatic ones. This is understandable because they are a pain in the ass to measure and not that heritable. But they have a huge effect on the individual. #4 I don’t think they had the data to correct for age properly, though that likely wouldn’t have bumped up the explanatory power by too much. #5 how many copies you have of the gene really matters #6 Gene x Environment interactions. If you eat a Mediterranean diet you’re a bit plump, a Japanese diet, you’re a bit thin. [In case it’s not clear, that’s a fake example of GxE] So hard to deal with this without experiments, but they didn’t really even try.
Side note: the functions of the identified SNPs is often to interfere with eating behavior, feelings of fullness, etc. So, while the genes they’ve identified don’t explain much variation, many of the effects are to make it harder to behave in ways that would make it easier to lose weight. I’m very curious about how much of the missing heritability affects similar pathways vs things like basic metabolism. That would really affect the treatment for people who are dangerously overweight.
This is a really lovely and readable summary paper on the current state of research and how it translates into clinical practice. It does a good job of not separating genetic and environment too much – it’s really impossible to totally isolate them.
You’re right of course. I was both exaggerating and ranting, which is kind of my modus operandi on here.
Obesity and its causes are multivariate, hard to quantify and hard to study accurately. I just bridle at the “genetics made me eat that whole cake and tub of ice cream” that much of the FA community now resorts to, and their increasing skinny shaming of women who for instance have cancer.
So that makes me more dogmatic than I should be. I prefer the real evidence, so thanks.
Anecdotally, my family’s genetics predisposes me to be heavy but I am not (any longer). So I have trouble with the FA “genetics made me do” it explanation personally, though population-level explanations are much different things.
Some interesting questions are even after the wide availability of processed foods in the 1950s and 1960s, obesity only really exploded in the late 80s and 90s. Was there some sort of threshold that food costs crossed? Some second-generation epistatic effect caused in part by chemical exposure? What? None of it makes much sense.
My hatred for the FA community and all such very American ideations makes me be less than objective at times. However, obviously fat people shouldn’t be shamed and telling people to have more willpower isn’t working.
Our whole food system should be reformed utterly, but unlike most liberals (not saying you specifically behave this way as I’ve seen no evidence of that at all) I believe also in personal responsibility as the other choice is (from my perspective) madness. Not the personal responsibility of conservatives where everything is blamed on the individual but rather some measure of personal responsibility that the FA crowd is all too willing to utterly abdicate — which is what makes the movement so very American.