Text

Physicist, Startup Founder, Blogger, Dad

Tuesday, May 14, 2013

Nature News: Chinese project probes the genetics of genius



This article is mostly correct -- see my comments below in [[ brackets ]]. As usual the Chinese connection is emphasized in the title, even though Plomin (Kings College London) is the more experienced researcher in this area, and most of our DNA samples come from US citizens.

To clarify, my main motivation for understanding the genetics of cognition derives from the observation that the human brain, the most complex object we know of in the universe, is produced from a genetic code of only gigabits in length. How, exactly, this works is one of the greatest scientific mysteries. Genomic selection and other "spin-offs" from this research are of secondary interest.
Nature News: The US adolescents who signed up for the Study of Mathematically Precocious Youth (SMPY) in the 1970s were the smartest of the smart, with mathematical and verbal-reasoning skills within the top 1% of the population. Now, researchers at BGI (formerly the Beijing Genomics Institute) in Shenzhen, China, the largest gene-sequencing facility in the world, are searching for the quirks of DNA that may contribute to such gifts. Plunging into an area that is littered with failures and riven with controversy, the researchers are scouring the genomes of 1,600 of these high-fliers in an ambitious project to find the first common genetic variants associated with human intelligence.

[[ SMPY qualifiers scored at the 1 in 10k level on the math portion of the SAT. Due to the positive correlation between M and V they almost all have V scores in the top half of one percent. ]]

The project, which was launched in August 2012 and is slated to begin data analysis in the next few months, has spawned wild accusations of eugenics plots, as well as more measured objections by social scientists who view such research as a distraction from pressing societal issues. Some geneticists, however, take issue with the study for a different reason. They say that it is highly unlikely to find anything of interest — because the sample size is too small and intelligence is too complex.

Earlier large studies with the same goal have failed. But scientists from BGI’s Cognitive Genomics group hope that their super-smart sample will give them an edge, because it should be enriched with bits of DNA that confer effects on intelligence. “An exceptional person gets you an order of magnitude more statistical power than if you took random people from the population — I’d say we have a fighting chance,” says Stephen Hsu, a theoretical physicist from Michigan State University in East Lansing, who acts as a scientific adviser to BGI and is one of the project’s leaders.

“If they think they’re likely to get much useful data out of this study, they’re almost certainly wrong,” says Daniel MacArthur, a geneticist at Massachusetts General Hospital in Boston. He is not against intelligence studies in principle, despite the visceral reactions they provoke in some people. “Studying intelligence is useful for understanding cognitive function, or diseases” that affect it, he says. But he questions whether the study will work.

[[ Not exactly sure what Dan means by "useful data" here. It's true that we don't anticipate getting more than a few genome-wide significant hits from a GWAS analysis. We may get zero! ]]

... Both Plomin and Hsu are passionate enough to take a shot, although their goals differ. Hsu is focused on the genetic basis of extreme intelligence. “My primary interest is why Einstein or Hawking is different from a normal person,” he says. Plomin is sequencing high-performers as a way of homing in on genes that affect intelligence in the broader population. If enough of these are discovered, he thinks that it may be possible to predict someone’s intelligence from an early age, and to offer help to children who are at risk of learning disabilities.

[[ This may give the false impression that it's a different genetic mechanism that gives rise to "extreme" intelligence as opposed to normal variation. ]]

Publicity around the project has spawned some extreme reactions. An article published in March entitled ‘China is Engineering Genius Babies’ in the US arts and culture magazine VICE branded the study “a state-endorsed genetic-engineering project” that will allow parents to predict the IQs of embryos and selectively breed ever-smarter children. (“That’s nuts,” says Hsu.) “Intelligence does push a lot of buttons. It’s like waving a red flag to a bull,” says Plomin. He argues that there is nothing wrong with using genetic information as the basis of educational interventions. “I’m interested in predicting learning problems early rather than waiting until kids get to school and then fail,” he says. ...

23 comments:

gwern said...

> Not exactly sure what Dan means by "useful data" here. It's true that we
don't anticipate getting more than a few genome-wide significant hits
from a GWAS analysis. We may get zero!

If you may get so few as zero, why are you running the project?

Emil Kirkegaard said...

Why is embryo selection nuts?

HughLygon said...

My Dad attended a university high. He was prez of his class. My Dad is an idiot and douche and, frankly, a source of shame for me. As I've said before, elite uni admissions ar ean obscenity. So are college loans. If Steve were a decent human being he'd leave the Satanic shit-hole that is American academia. He doesn't because he is a sociopath.

Christopher Chang said...

It's the "state-endorsed genetic engineering project" part that is nuts.

Emil Kirkegaard said...

What is nuts about that? The state has an interest in people getting smarter children, why not support genetic engineering products then? I don't support coercive eugenics if that's what is meant.

steve hsu said...

That's how science works. We don't know how many hits we'll get because we don't (yet) know what the underlying genetic architecture is. There were many failed GWAS before they found the first height hits at sample size of about 10k.

David Coughlin said...

I don't know any implementation of The State which is smarter than the smart people.

Richard Seiter said...

Can you offer any thoughts on if/how basing the GWAS on a full sequence rather than just SNPs might affect the power? Have there been any height GWAS run on full sequences?

steve hsu said...

I don't want to go into detail here, as this is a matter of ongoing research, but having whole genomes rather than SNPs should allow novel analyses that go beyond simple GWAS. For example, we may be able to say something about mutational load ... Of course I'm just an ignorant theoretical physicist, not a "real expert" on genomics ;-)

HughLygon said...

It would inevitably lead to Scandinavian socialism or "worse". The ruling neo-liberal ideology is opposed.

HughLygon said...

Then you know nothing.

nooffensebut said...

I stopped taking Daniel MacArthur seriously, and you should too.

“[T]he evidence for an association between the [MAOA] VNTR variant and antisocial behaviour is substantially more consistent than most of these associations. This may well be one of the rare cases of genuine associations.” --Daniel MacArthur on GNXP in 2009

“By historical analogy, most if not all of this [MAOA] literature is wrong, and will soon be forgotten.” --Daniel MacArthur on Genomes Unzipped in 2011

LaurentMelchiorTellier said...

First stature hit at n ~ 5k:
http://www.nature.com/ng/journal/v39/n10/abs/ng2121.html

(longitudinal - not case/control design!)

Then, replication of the result with larger sample sizes, leading to another 19 hits with n ~ 19k, and so on. Now, GIANT have hundreds of hits, after their dataset has become much larger.

BlackRoseML said...

And HMGA2 only explains around ~.3% and could be discovered in a GWAS involving 14,000 people at genome-wide significance. The latest GWAS by Plomin had a size of around 12,000, but they were studying youth (and IQ does not reach its maximum heritability then). If there were intelligence loci that explains > .5% of the phenotypic variance, they probably would have been found by now.

Dima Klondt said...

Enough with PR. Let's see the actual data.

steve hsu said...

I sympathize with your comment, but unless I refuse all contact from journalists this kind of thing is going to continue ...

LaurentMelchiorTellier said...

Analysis of the 5k led to discovery of two hits in HMGA2, which were targets for replication in the 14k.
Analysis of the 14k (or, rather, 19k) led to discovery of 20 hits, which were then again targets for replication. And so on.

Right, youth IQ has properties different to adult IQ.

NotaPhysicist said...

Is there any word on when individual data will be available from BGI? They said May.

BlackRoseML said...

A sample of 16.5k individuals was the replication sample in Weedon (2008) while 13.6k was size of the discovery sample.

LaurentMelchiorTellier said...

"To begin to identify genetic variants influencing height, we examined genome-wide association data from 4,921 individuals. Common variants in the HMGA2 oncogene, exemplified by rs1042725, were associated with height (P = 4 10-8). "

Though the original discovery sample is n = 5k (not 13.6k, you're looking at the second study from 2008, rather than the 2007 study linked above), it's a fair point that you can count the replication sample in different ways. Strictly speaking, they replicate HMGA2 in 29k in the original Weedon, and at higher n the next year. Even more strictly, they replicate in much larger n, in ensuing studies. And so on. Each confirmatory GIANT collation can be described as a replication.

HughLygon said...

Too bad a lot of those kids are white gentiles.

corporate_serf said...

How do they measure extreme mathematical giftedness? Surely SAT scores cannot distinguish extreme intelligence from moderate? Perhaps performance at math olympiads is a better choice, but talking to some friends whose offspring are going through this process, it seems that schools (at least in the US) has dumbed even that down, in order to let more people put something on the resume

Christopher Chang said...

We're starting the data return process now. You should receive an email within a few days, and (after optionally posting a GPG public key which we'll use for encryption) your data will be downloadable shortly afterwards.

Blog Archive

Labels