The following claim is being made by Timothy Frayling of Peninsula Medical School in Exeter, UK, and colleagues:

A genetic survey of more than 34,000 people has revealed the first gene known to have a decisive effect on height in people of average stature. A change to just a single letter of genetic code is linked to a height boost of almost a centimetre in a healthy person, all other things being equal.

Although up to 90% of variation in people’s height is thought to be down to genetics, identifying the genes involved is difficult because there are thought to be hundreds of them, each with an almost imperceptible effect.

[source]

The real reason to study height: so you can genetically modify your kid so s/he can perform in high school sports. But this photograph is misleading. The research discussed here shows how a gene may cause variation in height that is less than the thickness of these athlete’s innersole inserts. The difference you see in height in this photograph has nothing to do with this research.

Instant reality check: Let me say unequivocally that it is not “thought” that 90% of human variation in stature is genetic, just in case you thought that from the above. Note that the sentence says “up to” which includes 0%, and that it does not give any basis for this belief. In truth, almost no variation in human height in “normal” groups has been accounted for by genetic variation. Indeed, the vast majority of known variation in human stature that has actually been measured is the very large difference … in the two-digit percent range … that occurs in so-called “secular variation.” An example of this is the well known increase in height of various populations around the world over decades of time, following immigration events. There is simply no way that this variation is genetic. We have no evidence that the short people were being killed off and/or the tall people surviving or mating more, etc.

The idea that there are “hundreds of genes” is also worth further investigation. Given that there are not any known, why would the presumption be made that there are “hundreds” of “genes for stature?” There are two reasons for this, one bogus, the other tricky. First the bogus one. Human stature is cited in all textbooks that deal with this topic as a classic example of “continuous genetic variation.” Since genes and their direct effects are rarely continuous, continuity in phenotype would then be explained by there being several interacting genes. Why is that?

Let’s say you are short of cash so you go out on the street and start panhandling. If you panhandle long enough to get one person to give you some money, it is very unlikely that you have 23 cents. Most likely, you have 5, 10, 25 cents, some combination of those, or if you are lucky, you have a dollar.

If you panhandle for much longer, you could have any one of several amounts, but most likely the amount will end in a zero or a five, because you’ve got to be kind of an asshole to give a panhandler pennies, and there are not that many assholes.

But if you panhandle for a really really long time, eventually you are going to stop at Dunn Brothers to get a Grande Latte or something and end up with pennies in your change, or somebody is going to give you a couple of pennies along with a few nickels and dimes. Given enough time … enough different sources contributing to the money in your pocket … you can expect that you will have any one of a large, continuous, smoothly distributed set of amounts.

So there must be many, many genes involved in height because height varies along a continuum. It’s not like wrinkled vs. smooth peas.

This, of course, is utterly stupid, even though it is in all the textbooks. Even if there were a small number of genes involved in stature in typical human populations, there has to be a fair amount of norm of reaction (phenotypic variation that is caused by interaction with the environment), which would easily smooth out stature “amounts” that would otherwise be discrete.

So that was the bogus reason that we often here about the “many many” genes involved in stature. The “tricky” reason is this: What is stature? It is the manifestation of growth with a little shape on the side. Most studies of mammals actually don’t deal that much with stature, but usually use mass, but mass in humans is trickier than stature so stature is how humans are generally compared. This is a big pain for those of us who try to compare humans with other mammals. First, humans went and evolved this whole “upright on two legs” thing so stature/mass/locomotion etc. are all screwed up, then human-oriented scientists have often collected stature data but not mass data!

Anyway, stature reflects growth. Growth is one of those trade-off things. Imagine two identical twins, growing. One of them falls ill with various infections several times over a three year period while growing, the other does not. Otherwise they are the same … they do the same things, they eat the same things, etc. The one who got sick will end up with a completed stature (adult height) lower than the one who did not get sick, because energy that otherwise would have been used for growth was instead used for the immune responses to the infections.

What if the sickly twin was actually not genetically identical to the other twin, but in fact, had a mutation that caused these frequent infections to occur? This would be a mutation making her more susceptible to initially getting the infection, but everything else … response to the infection and other aspects of the individual’s physiology … are the same.

This would be a gene for stature because it is a gene, it varies, and its variation determines a part of the variation in stature.

If you discovered this gene in a large scale genetic study looking at stature, you might wonder about what else the gene does, and you might even figure that out eventually, but on initial investigation you would have a gene for stature in hand.

There are, therefore, a zillion genes that affect stature, because stature is the mirror of growth process, and growth is always involved in a trade off against other processes. So any gene that effects any aspect of energy allocation is a gene for stature.

Let’s look back at the report in Nature:

Researchers therefore combed through almost the entire genome of nearly 5,000 volunteers in search of tiny changes, called polymorphisms, that correspond to variations in height. Eventually they found a single-letter DNA substitution, buried in a gene called HMGA2, that influences height.

Ding ding ding ding ding…..

That was my bullshit bell going off. I’m not saying that this is bullshit. Sometimes the bullshit bell goes off when it is not needed. It’s a little sensitive.

How many genes are there in this genome database? I’m going to give them a break and say 10,000. You get 5,000 cases distributed across 10,000 independent variables (genes) and you are looking for any amount (in any direction) of change in one dependent variable (stature). Holy crap. They only found one???? At first glance, this should make you worry.

But don’t worry. I think they did not screw this up. But I wanted to point out why the bullshit bell would go off here. If you look at many variables, some will, by random chance alone, appear to “work” in your study. You’ve got to do some other stuff beyond this kind of fishing expedition. And these researchers did just that:

Follow-up testing of some 29,000 people confirmed that HMGA2 does indeed affect height…

OK, that was good. They probably had many, many “hits” in the first study and then ruled most of them out with the follow up. It is still possible that this is an accidental linkage, but much less likely.

But, let’s look at the effect they are describing here, what they claim the “gene” is “doing.” Statistics are of no interest whatsoever unless the phenomenon makes “biological sense” in some way.

People with two copies of the ‘tall’ variant of HMGA2 are, on average, almost a centimetre taller than those with two copies of the ’short’ version. Those with one copy of each are somewhere in the middle

That there is a pattern like the one normally expected with a simple genetic system is comforting … this may indeed be a gene that has an influence on height. But notice that we are talking about “almost a centimetre” … a very very small effect.

If we think of this from an adaptationist perspective … that there are multiple alleles, and that these alleles are under differential selection … then I ask you: How is less than one centimeter …. that’s like the thickness of your earlobe …. going to be selected for or against? This amount of stature can’t possibly be under selection. But something else might be under selection. MIght.

What would that be? See above. Something having to do with the immune system, or whatever. Something that takes a bit away from growth, or otherwise affects overall physiological process.

The researchers have a suggestion.

It’s also not exactly clear how HMGA2 influences height, despite the fact that rare severe mutations in the gene are already known to alter body size in both mice and humans. The gene is involved in unravelling the protein-rich structure, called chromatin, in which chromosomes are packaged, leading Frayling to speculate that the gene may influence the speed at which DNA is replicated during cell division, thus affecting overall body growth.

Ding ding ding ding ding…..

Oh, crap, the bullshit meter is going off again. Maybe it just wants a new battery. Let’s see.

OK, the BS-M went off regarding this mention of “severe mutations in the gene … already known to alter body size in both mice and humans”

The problem with this is simply that severe mutations, i.e., a really really screwed up allele, like one that does not produce a functional protein of any kind, can really really screw up a phenotype in ways that may be very indirectly related to, or not at all related to, the system you are looking at. I note that the author of this report is being cautious in this paragraph, so the BS-M was probably being over sensitive.

The meat of this paragraph is very interesting. This gene may code for a protein involved in unraveling of chromatin. In other words, it has to do with mitosis and/or meiosis (cell division). Mitosis, it’s rate and timing, is fundamental to a wide range of physiological processes and systems, including growth.

So does this all make biological sense? Not yet.

Cells go through certain cycles known as S, G1, M and G2. The S phase is when the chromosomes are replicated. This phase is initiated by a protein called Cyclin-A. Our gene, the HMGA-2 gene, appears to influence, perhaps even cause, Cyclin-A to be expressed. So it’s pretty basic in the cell cycle. Something so basic can only have an influence on something so “final” as height in a fairly indirect way.

Nevertheless, simply knowing which genes help to determine height could help doctors decide whether small kids have naturally ’short’ genes, or whether they are suffering from a medical condition that stunts growth. “For a lot of children who perhaps are a bit behind their growth chart, there is a pressure for doctors to treat them or find something wrong with them,” Frayling says. “If we can find 50 or 100 height genes, we could look at them and say ‘this is entirely in keeping with your height profile’.”

Ding … ding … ZCKL-eerrk

OK, I’m going to turn off the BS-M right now because from here on out it’s not going to stop running. This is not really about the research. Finding out how variation in a gene relates to variation in phenotype is basic stuff and I’m glad they are doing it. But the rest of the report in Nature is annoying.

In the above quote we see an adherence to the idea that stature is polygenic. But stature is not really a trait in this sense, it is one aspect of a set of competing processes. Each pathway relating a gene to stature is going to be involved in so many other processes that one will not really be able to conclude anything useful form this kind of analysis.

In the meantime, however, Frayling and his colleagues are searching for yet more genes that influence height. “We won’t expose all of the genetic basis of height, but over the next couple of years, we might find several hundred [genes] — perhaps 50% of the variation,” Frayling says.

More of the same. It’s called the “Job Preservation Strategy.” This is Part A of a two part strategy for job preservation: Show that this work you have been doing over the last year or so, with an NIH grant, needs to be done fifty more times, with more grants.

Again, I do not begrudge these researchers this funding, and again, I think this is important work. And everybody else does this. Every time a first responder pulls a child out of a burning building, everybody realizes how important first responders are. Scientists need their chance to seem important as well. Part A is just fine withe me. It’s really Part B of the strategy that annoys me most, as demonstrated here:

The discovery could also help to piece together the genetic framework of diseases such as diabetes and cancer, suggests another member of the research group, Joel Hirschhorn of the MIT Broad Institute and the Children’s Hospital, both in Boston, Massachusetts.

There is some evidence that slightly shorter-than-average people are susceptible to diabetes, although this may be for socio-economic reasons. Conversely, taller people tend to be more prone to cancer, perhaps because they simply have more growing cells in which the disease can arise.

Discovering the genes that govern height would allow geneticists to examine whether different versions of these genes are linked to increased rates of various diseases, and could even help them work out the mechanisms involved.

Part B is to link your research to cancer, diabetes, sick children, and so on. That’s good if its for real. Actually, our gene, HMBA-2, is linked to tumor growth. It seems to be more abundant than it “should” be in some tumors. One could ask, then, why are these researchers dealing with stature and not tumors?

The reason? Because the link between allelic variation and stature is interesting. In my view, that should be enough. But it is not enough because we always have to be framing things. So, holding up the sick baby and pointing to its tumor is Part B of the strategy.

In conclusion: If 90 % of “normal” human stature variation in “typical” populations is genetic, why do we only know about this one gene that accounts for less variation than the thickness of the sports section in the daily paper?

[source]

_________________

Caveat: There are genes related to stature in “normal human populations” but they are special cases that have nothing interesting to do with the present discussion.