Scientists are finding more genes linked to IQ. This doesn’t mean we can predict intelligence. – Vox

Last month, researchers announced some astonishing findings in Nature Genetics: Theyd found 40 genes that play a role in shaping human intelligence, bringing the total number of known intelligence genes up to 52.

This study was a big deal because while weve known intelligence is largely heritable, we havent understood the specifics of the biology of IQ why it can be so different between people, and why we can lose it near the end of life.

The Nature Genetics study was a key early step toward understanding this, hailed as an enormous success in the New York Times.

And there are many more insights like this to come. The researchers used a design called a genome-wide association study. In it, computers comb through enormous data sets of human genomes to find variations among them that point to disease or traits like intelligence. As more people have their genomes sequenced, and as computers become more sophisticated at seeking out patterns in data, these types of studies will proliferate.

But theres also a deep uneasiness at the heart of this research it is easily misused by people who want to make claims about racial superiority and differences between groups. Such concerns prompted Nature to run an editorial stressing that the new science of genetics and intelligence comes to no such conclusions. Environment is crucial, too, Nature emphasized. The existence of genes for intelligence would not imply that education is wasted on people without those genes. Geneticists burned down that straw man long ago.

Also, nothing in this work suggests there are genetic difference in intelligence when comparing people of different ancestries. If anything, it suggests that the genetics that give rise to IQ are more subtle and intricate than we can ever really understand.

Were going to keep getting better at mapping the genes that make us smart, make us sick, or even make us lose our hair. But old fears and myths about genetics and determinism will rear their heads. So will fears about mapping ideal human genes that will lead to designer babies, where parents can pick traits for their children la carte.

To walk through the science, and to bust its myths, I spoke to Danielle Posthuma, a statistical geneticist at Vrije Universiteit in Amsterdam, who was the senior author on the latest Nature study.

Theres a simple understanding of genetics were all taught in high school. We learn, as Gregor Mendel discovered with pea plants, that we can inherit multiple forms of the same gene. One variation of the gene makes wrinkled peas; the other makes for round peas. Its true, but its hardly the whole story.

In humans, a few traits and illnesses work like this. Whether the bottom of your earlobes stick to the side of your face or hang free is the result of one gene. Huntingtons disease which deteriorates nerve cells in the brain is the result of a single gene.

But most of the traits that make you you your height, your personality, your intellect arise out of a complex constellation of genes. There might be 1,000 genes that influence intelligence, for example. Same goes for the genes that lead to certain disorders. Theres no one gene for schizophrenia, for obesity, for depression.

A single gene for one of these things also wont have an appreciable impact on behavior. If you have the bad variant of one gene for IQ, maybe your IQ score ... is 0.001 percent lower than it would have been, Posthuma says.

But if you have 100 bad variants, or 1,000, then that might make a meaningful difference.

Genome-wide association studies allow scientists to start to see how combinations of many, many genes interact in complicated ways. And it takes huge data sets to sort through all the genetic noise and find variants that truly make a difference on traits like intelligence.

The researchers had one: the UK Biobank, a library that contains genetic, health, and behavioral information on 500,000 Britons. For the study, they pulled complete genome information on 78,000 individuals who had also undergone intelligence testing. Then a computer program combed through millions of sites on the gene code where people tend to variate from one another, and singled out the areas that correlated with smarts.

The computer processing power needed for this kind of research this study had to crunch 9.3 million DNA letters from 78,000 people hasnt been available very long. But now that it is, researchers have been starting to piece together the puzzle that links genes to behaviors.

A recent genome-wide analysis effort identified 250 gene sites that predicted male pattern baldness in a sample of 52,000 men. (Would you really want to know if you had them?) And theres been progress identifying genes that signal risk for diabetes, schizophrenia, and depression.

And these studies dont just look at traits, diseases, and behavior. Theyre also starting to analyze genetic associations to life outcomes. A 2016 paper in Nature reported on 74 gene sites that correlate with educational attainment. (These genes, the study authors note, seem to have something to do with the formation of neurons.) Again, these associations are tiny the study found that these 74 gene variants could only explain 3 percent of the difference between any two people on what level of education they achieve. Its hardly set in stone that youll flunk school if you dont have these gene variants.

But still, they make a small significant difference once you start looking at huge numbers of people.

Its important to note that Posthumas study was only on people of European ancestry. Whatever we find for Europeans doesnt necessarily [extrapolate] for Asians or South Americans, [or any other group] she says. Those things are often misused.

Which is to say: The gene variations that produce the differences between Europeans arent necessarily the same variations that produce differences among groups of different ancestry. So if you were to test the DNA of someone of African origin, and saw they lacked these genes, it would be incredibly irresponsible to conclude they had a lower capacity for intelligence. (Again, there are also likely hundreds of more genetic sites that have something to do with intellect that have yet to be discovered.)

Posthumas work identifying genes associated with intelligence isnt about making predictions about how smart a baby might grow up to be. She doesnt think you can reliably predict educational or intelligence outcomes from DNA alone. This is all really about reverse-engineering the biology of intelligence.

Genes code for proteins. Proteins then interact with other proteins. Researchers can trace this pathway all the way up to the level of behavior. And somewhere along that path, there just might be a place where we can intervene and stop age-related cognitive decline, for instance, and Alzheimers.

We're finally starting to see robust reliable associations from genes with their behavior, she says. The next step is how do we prove that this gene is actually evolved in a disorder, and how does it work?

Understanding the biology of intelligence could also lead the way for personalized approaches to treating neurodegenerative diseases. Its possible that two people with Alzheimers may have different underlying genetic causes. Knowing which genes are causing the disease, then, you might be able to tailor the treatment, Posthuma says.

As more and more genome-wide studies are conducted, the more researchers will be able to assign people polygenic risk scores for how susceptible they might be for certain traits and diseases. That can lead to early interventions. (Or, perhaps in the wrong hands, a cruel and unfair sorting of society. Have you seen the movie Gattaca?)

And there are some worries about abusing this data, especially as more and more people get their genomes analyzed by commercial companies like 23&Me.

Many people are concerned that insurance companies will use it, she says. That they will look into people's DNA and say, Well, you have a very high risk of being a nicotine addict. So we want you to pay more. Or, You have a high risk of dying early from cancer. So you have to pay more early in life. And of course, that's all nonsense. Its still too complicated to make such precise predictions.

We now have powerful tools to edit genes. CRISPR/Cas9 makes it possible to cut out any specific gene and replace it with another. Genetic engineering has advanced to the point where scientists are building whole organisms from the ground up with custom DNA.

Its easy to indulge our imaginations here: Genome-wide studies are going to make it easier to predict what set of genes leads to certain life outcomes. Genetic engineering is making it easier to assemble whatever genes we want in an individual. Is this the perfect recipe for designer babies?

Posthuma urges caution here, and says this conclusion is far afield from the actual state of the research.

Lets say you wanted to design a human with superior intelligence. Could you just select the right variants of the 52 intelligence genes, and wham-o, we have our next Einstein?

No. Genetics is so, so much more complicated than that.

For one, there could be thousands of genes that influence intelligence that have yet to be discovered. And they interact with each other in unpredictable ways. A gene that increases your smarts could also increase your risk for schizophrenia. Or change some other trait slightly. There are trade-offs and feedback loops everywhere you look in the genome.

If you would have to start constructing a human being from scratch, and you would have to build in all these little effects, I think we wouldn't be able to do that, Posthuma says. It's very difficult to understand the dynamics.

There are about 20,000 human genes, made up of around 3 billion base pairs. We will never be able to fully predict how a person will turn out based on the DNA, she says. Its just too intricate, too complicated, and also influenced heavily by our environment.

So you could have a very high liability for depression, but it will only happen if you go through a divorce, she says. And who can predict that?

And, Posthuma cautions, there are some things that genome-wide studies cant do. They cant, for instance, find very, very rare gene variations. (Think about it: If one person in 50,000 has a gene that causes a disease, its just going to look like noise.) For schizophrenia, she says, we know that there's some [gene] variants that decrease or increase your risk of schizophrenia 20-fold, but they're very rare in the population.

And they cant be used to make generalizations about differences between large groups of people.

Last year, I interviewed Paul Glimcher, a New York University social scientist whose research floored me. Glimcher plans to recruit 10,000 New Yorkers and track everything about them for decades. Everything: full genome data, medical records, diet, credit card transactions, physical activity, personality test scores, you name it. The idea, he says, is to create a dense, longitudinal database of human life that machine learning programs can mine for insights. Its possible this approach will elucidate the complex interactions of genetics, behavior, and environment that put us at risk for diseases like Alzheimers.

Computer science and biology are converging to make these audacious projects easier. And to some degree, the results of these projects may help us align our genes and our environments for optimal well-being.

Again, Posthuma cautions: Not all the predictions this research makes will be meaningful.

Do we care if we find a gene that only increases our height or our BMI or our intelligence with less than 0.0001 percent? she asks. It doesn't have any clinical relevance. But it will aid our scientific understanding of how intellect arises nonetheless.

And thats the bottom line. The scientists doing this work arent in it to become fortune tellers. Theyre in it to understand basic science.

What most people focus on, when they hear about genes for IQ, they say: Oh, no. You can look at my DNA. You can tell me what my IQ score will be, Posthuma says. They probably dont know its much better if you just take the IQ test. Much faster.

Read the original here:
Scientists are finding more genes linked to IQ. This doesn't mean we can predict intelligence. - Vox