Genetic Predictions - Height Edition

[jaa]

Stephen Tsu over at Information Processing just posted his team's latest results on predicting height. They used machine learning algorithms on ~500k genomes which uses info from thousands of SNPs. The results are they can predict height to within a few centimetres for any individual.

Information Processing: Accurate Genomic Prediction Of Human Height

The big picture implication is that heritable complex traits controlled by thousands of genetic loci can, with enough data and analysis, be predicted from DNA. I expect that with good genotype | phenotype data from a million individuals we could achieve similar success with cognitive ability. We've also analyzed the sample size requirements for disease risk prediction, and they are similar (i.e., ~100 times sparsity of the effects vector; so ~100k cases + controls for a condition affected by ~1000 loci).

Disease, IQ, and height are all similar. All have polygenic contributors. This further increases my confidence that more data + better analysers will continue to make the strong case for genetic influence (say >50% environment) in these areas. Arguments about the failure of genetics seem dead in the water.

 

[MatheusPN]

If Haidut, counter-argument, would be interesting, because there always appears a letter in his sleeve

 

[jb116]

Stephen Tsu over at Information Processing just posted his team's latest results on predicting height. They used machine learning algorithms on ~500k genomes which uses info from thousands of SNPs. The results are they can predict height to within a few centimetres for any individual.

Information Processing: Accurate Genomic Prediction Of Human Height



Disease, IQ, and height are all similar. All have polygenic contributors. This further increases my confidence that more data + better analysers will continue to make the strong case for genetic influence (say >50% environment) in these areas. Arguments about the failure of genetics seem dead in the water.

" Arguments about the failure of genetics seem dead in the water."


Not at all. Why do you say that?

 

[haidut]

Stephen Tsu over at Information Processing just posted his team's latest results on predicting height. They used machine learning algorithms on ~500k genomes which uses info from thousands of SNPs. The results are they can predict height to within a few centimetres for any individual.

Information Processing: Accurate Genomic Prediction Of Human Height



Disease, IQ, and height are all similar. All have polygenic contributors. This further increases my confidence that more data + better analysers will continue to make the strong case for genetic influence (say >50% environment) in these areas. Arguments about the failure of genetics seem dead in the water.

They must have missed the news about the study in humans showing it is protein quality, not genetics that determines height.
Protein Quality, Not Genes, Determine Male Height

 

[haidut]

If Haidut, counter-argument, would be interesting, because there always appears a letter in his sleeve

Well, there is this thread I posted, which seems to show the exact opposite - in humans. I think @jb116 will probably find it interesting as well.
Protein Quality, Not Genes, Determine Male Height

 

[jb116]

" Arguments about the failure of genetics seem dead in the water."


Not at all. Why do you say that?

I think this needs to be put into better perspective. I for one don't see this as a vindication of "genetics rules all." It all depends on how far your sight goes, so to speak.
When you look at geneticism through a narrow window, you see genes as a cause. When you consider the generational spread, you see genes as a layout of inheritance; as "effects."
Haidut's studies demonstrate that idea: seeing past the sliver of gene theory and understanding that they actually illustrate great changes from without, including diet and environment.

 

[Mukem]

Well, there is this thread I posted, which seems to show the exact opposite - in humans. I think @jb116 will probably find it interesting as well.
Protein Quality, Not Genes, Determine Male Height

Does that study demonstrate anything but an association? You say the people in the mountains have the same genes, but how do you know?
As far as I can tell, the original study even claims genes to be responsible for the stature of the mountain men.

 

[jb116]

Does that study demonstrate anything but an association? You say the people in the mountains have the same genes, but how do you know?
As far as I can tell, the original study even claims genes to be responsible for the stature of the mountain men.

Mukem, see my above post.
It's important you understand the idea other wise you'll be asking that question in circles.

 

[CarbAppreciator]

A couple centimeters doesn't seem very accurate.
There is definitely a genetic component, but genetics are extremely pliable even within 3 generations, all the epigentics science shows this. haiduts post is the key, if one has tall genetics, and doesn't get good protein, they will be short: Environmental factors are always causal.

 

[Mukem]

Mukem, see my above post.
It's important you understand the idea other wise you'll be asking that question in circles.

What's your point? That I'm not allowed to think genes may have an effect? Maybe I read all these posts wrong but I haven't seen any evidence of a causal role of either certain genes or certain environmental factors, only associations.

 

[jb116]

What's your point? That I'm not allowed to think genes may have an effect? Maybe I read all these posts wrong but I haven't seen any evidence of a causal role of either certain genes or certain environmental factors, only associations.

My point is to think outside the box a bit. Put aside the studies and just think. You know the best evidence that genes do not rule all and environment is the ultimate factor?
Think of individuals who have a condition NOBODY in their family/family line has ever had. You see that more and more these days in fact. It's simple evidence for it.
The extreme anti-metabolic conditions individuals are exposed to now can create novel disease conditions. Genetics are a result and culmination of generational environmental and diet factors.

 

[Mukem]

My point is to think outside the box a bit. Put aside the studies and just think. You know the best evidence that genes do not rule all and environment is the ultimate factor?
Think of individuals who have a condition NOBODY in their family/family line has ever had. You see that more and more these days in fact. It's simple evidence for it.
The extreme anti-metabolic conditions individuals are exposed to now can create novel disease conditions. Genetics are a result and culmination of generational environmental and diet factors.

I don't get how it's evidence for anything, to be honest. That genetics are a result is just your opinion.

 

[jb116]

I don't get how it's evidence for anything, to be honest. That genetics are a result is just your opinion.

The fact that a person has a disease without the influence of genetics or any sign of such disease in their family, I think is pretty damning evidence that genes too are an effect than a cause.
There is no other way to side-step that. What else would influence a disease state if genes rule and yet there is no genetic indication for such disease in a given individual.

 

[Mukem]

It may be that the person ended up diseased due to some environmental factor and not due to genetic interactions (it's not certain), which would show environment can play a role. In height, some environmental factors may also play a role. But I haven't seen clear evidence that height is determined by the environment. Aside from people who think that genes determine everything and people who think that genes do nothing there are also a lot of people in between.

 

[jb116]

It may be that the person ended up diseased due to some environmental factor and not due to genetic interactions (it's not certain), which would show environment can play a role. In height, some environmental factors may also play a role. But I haven't seen clear evidence that height is determined by the environment. Aside from people who think that genes determine everything and people who think that genes do nothing there are also a lot of people in between.

I don't think any of us think genes do nothing per se, at least I can speak for myself. Since effects can have an effect, genes are no exception there. The point is, the OP does not show that arguments of the failure of genes are dead in the water. I think you understand that as well, since you concede "diseased due to some environmental factor." The evidence is still simply clear that genes are not the determining, ruling thing they are made out to be otherwise the clean familial history of an individual who is diseased would and should have determined that said individual's fate in an absolute sense.

 

[Peater Piper]

A couple centimeters doesn't seem very accurate.

Then men in my family range from 5'8" to 6'8". The women, 5'2" to 6'2". That's a 30 cm range. If a genetic study accurately predicted their height within 2 cm, I'd be highly impressed. Looking at the chart in the original link though, there were some significant outliers. It looks like some people ended up 10 cm above or below the prediction, although the plot shows the majority of the results were much closer than this, and there's a very clear trend. Considering their height predictions had about a 40 cm range, and none missed by more than about 10 cm (with the majority being much closer to that), I'd say that points to a very strong role of genetics. No one was predicted to be 150 cm who came anywhere close to 190 cm, and vice versa.

Regarding nutrition, I don't think anyone here doubts that a hypocaloric, low protein diet can stunt growth. However, having grown up with a brother who grew to be much larger than I am, and spending a lot of time with family members who are way above the average height of anyone in the world, I can say that there's nothing particularly impressive about what they eat or the amounts. Actually, in the case of several of them, I'd suggest their protein intake was quite low. Very little meat or dairy, a lot of junk food. It's shocking to be honest.

 

[haidut]

To @Mukem and @jb116 - below is yet another fascinating article on the importance of environment over genes. And since the studies that article discusses used animals they were interventional and not just an association. So, the superiority of environment over genetics when it comes to height, and (just as importantly) the ability to pass on these non-genomic changes in height to offspring are on full display once again, and in many different species. And apparently, the effects of low-protein diet extend beyond just height.
http://nautil.us/issue/58/self/heredity-beyond-the-gene

"...But is any of this enormous environmentally induced variation in male phenotype transmitted across generations? To find out, we generated variation in male body size by rearing some larvae on a nutrient-rich larval medium, while rearing their siblings on a diluted medium. This resulted in sets of large and small brothers, which we then paired with females that had all been reared on the same larval food. When we measured the offspring, we found that large males produced larger offspring than their small brothers, and subsequent work showed that this nongenetic paternal effect is probably mediated by substances transferred in the seminal fluid.34,35However, because T. angusticollis males transfer a tiny ejaculate, orders of magnitude smaller than the typical nutrient-laden ejaculates produced by males of some insect species, this effect does not appear to involve the transfer of nutrients from males to females or to their offspring."

"...There is ample evidence that parental diet can affect offspring in mammals as well. Experimental research on the effects of diet in rats—particularly the restriction of key nutrients such as protein—began in the first half of the 20th century with the objective of gaining insight into the health consequences of malnutrition. In the 1960s, researchers were intrigued to discover that female rats fed a low-protein diet during pregnancy produced offspring and grand-offspring that were sickly and scrawny and had relatively small brains with a reduced number of neurons, scoring poorly on tests of intelligence and memory. In recent years, research efforts have turned to understanding the effects of excessive or unbalanced nutrient intake, using rats and mice as experimental models to gain insight into the human obesity epidemic, and it is now well-established that both maternal and paternal diets can have a variety of effects on offspring development and health. Some of these effects come about via epigenetic reprogramming of embryonic stem cells in the womb. For example, in rats, a high-fat maternal diet has been shown to reduce the proliferation of hematopoietic stem cells (which give rise to blood cells), while a maternal diet rich in methyl-donors has been found to promote the proliferation of neuronal stem cells in embryos.37,38 In rats, a high-fat paternal diet has been found to cause reduced insulin secretion and glucose tolerance in daughters.39 There is evidence of such effects in humans as well."

"...Stepping back to survey the current state of knowledge in the study of extended heredity, we are reminded of genetics in the 1920s or molecular biology in the 1950s. We know just enough to fathom the depths of our ignorance and to recognize the challenges that lie ahead. But one conclusion that is already beyond reasonable doubt is that the Galtonian assumptions that have shaped both empirical and theoretical research for nearly a century are violated in many contexts, and this means that biology has exciting times before it. Empirical researchers will be busy for many years exploring the mechanisms of nongenetic inheritance, observing its ecological effects, and establishing its evolutionary consequences. This work will require developing new tools and devising ingenious experiments. Theoreticians have the equally important task of clarifying ideas and generating predictions. And on a practical level, in medicine and public health, it is now equally clear that we need not be “passive transmitters of a nature we have received,” because our life experiences play a nontrivial role in shaping the hereditary “nature” that we transmit to our children."

 

[schultz]

Aren't the genes always being altered and/or deranged by the environment? If genes can be "switched on and off" by the environment (food, stress, radiation, etc.) then things can quickly change from one generation to another and even during the organisms life. Do I have this right?

Forgive my ignorance as it's not a topic I know much about, I've just heard Ray talk about it on the various podcasts, but it can be hard to remember things when you don't fully understand what Ray is talking about.

 

[Mukem]

Thanks, that's interesting. It still does not show superiority of environment over genes, only that the environment has effects, eg. on size of larvae (extrapolating this to human height is a bit of a leap). The non-genomic inheritance is interesting too.

 

[jb116]

To @Mukem and @jb116 - below is yet another fascinating article on the importance of environment over genes. And since the studies that article discusses used animals they were interventional and not just an association. So, the superiority of environment over genetics when it comes to height, and (just as importantly) the ability to pass on these non-genomic changes in height to offspring are on full display once again, and in many different species. And apparently, the effects of low-protein diet extend beyond just height.
http://nautil.us/issue/58/self/heredity-beyond-the-gene

"...But is any of this enormous environmentally induced variation in male phenotype transmitted across generations? To find out, we generated variation in male body size by rearing some larvae on a nutrient-rich larval medium, while rearing their siblings on a diluted medium. This resulted in sets of large and small brothers, which we then paired with females that had all been reared on the same larval food. When we measured the offspring, we found that large males produced larger offspring than their small brothers, and subsequent work showed that this nongenetic paternal effect is probably mediated by substances transferred in the seminal fluid.34,35However, because T. angusticollis males transfer a tiny ejaculate, orders of magnitude smaller than the typical nutrient-laden ejaculates produced by males of some insect species, this effect does not appear to involve the transfer of nutrients from males to females or to their offspring."

"...There is ample evidence that parental diet can affect offspring in mammals as well. Experimental research on the effects of diet in rats—particularly the restriction of key nutrients such as protein—began in the first half of the 20th century with the objective of gaining insight into the health consequences of malnutrition. In the 1960s, researchers were intrigued to discover that female rats fed a low-protein diet during pregnancy produced offspring and grand-offspring that were sickly and scrawny and had relatively small brains with a reduced number of neurons, scoring poorly on tests of intelligence and memory. In recent years, research efforts have turned to understanding the effects of excessive or unbalanced nutrient intake, using rats and mice as experimental models to gain insight into the human obesity epidemic, and it is now well-established that both maternal and paternal diets can have a variety of effects on offspring development and health. Some of these effects come about via epigenetic reprogramming of embryonic stem cells in the womb. For example, in rats, a high-fat maternal diet has been shown to reduce the proliferation of hematopoietic stem cells (which give rise to blood cells), while a maternal diet rich in methyl-donors has been found to promote the proliferation of neuronal stem cells in embryos.37,38 In rats, a high-fat paternal diet has been found to cause reduced insulin secretion and glucose tolerance in daughters.39 There is evidence of such effects in humans as well."

"...Stepping back to survey the current state of knowledge in the study of extended heredity, we are reminded of genetics in the 1920s or molecular biology in the 1950s. We know just enough to fathom the depths of our ignorance and to recognize the challenges that lie ahead. But one conclusion that is already beyond reasonable doubt is that the Galtonian assumptions that have shaped both empirical and theoretical research for nearly a century are violated in many contexts, and this means that biology has exciting times before it. Empirical researchers will be busy for many years exploring the mechanisms of nongenetic inheritance, observing its ecological effects, and establishing its evolutionary consequences. This work will require developing new tools and devising ingenious experiments. Theoreticians have the equally important task of clarifying ideas and generating predictions. And on a practical level, in medicine and public health, it is now equally clear that we need not be “passive transmitters of a nature we have received,” because our life experiences play a nontrivial role in shaping the hereditary “nature” that we transmit to our children."

Thanks! I think this exactly goes hand in hand with my point about anomalies of disease states found in individuals that otherwise are not found generationally. It tells you that the environment and diet is indeed severely shaping the outcome of a person's health state.