Just how well could you design a baby?
Study questions whether IQ or height can be predicted at all.
By Paul Biegler
Not everyone wants to raise the lovechild of Albert Einstein and Arnold Schwarzenegger but, like it or not, designer babies are inching their way into the global marketplace.
This month it was reported that US start-up Genomic Predictions is offering genetic testing of IVF embryos that includes, among others, measures of intelligence and height.
The move reignites an ethical firestorm on predictive genetic testing.
Philosophers such as Julian Savulescu, from the University of Oxford, have argued parents have a moral obligation to have the “best” possible child.
But a new study, published in the journal Cell, may render much of the kerfuffle moot. At least for now.
Led by Todd Lencz at the Feinstein Institutes of Medical Research in New York, US, the study used modelling and real-world outcomes to question whether IQ or height can be predicted at all.
The team took a hard look at genome wide association studies (GWAS) that link the genetic makeup of hundreds of thousands of people with their IQ and height. They wanted to get a handle on what gene patterns might predict the traits.
It’s a many-headed beast because, unlike disorders such as cystic fibrosis that are caused by mutations in a single gene, intelligence and height are “polygenic” – determined by many genes.
Armed with software and algorithms sufficiently powerful to crunch the numbers, Lencz’s team used the data to predict height and IQ in a bunch of “simulated embryos”.
This did not require simulated sex. Rather the virtual offspring came from pairing up the genetic data of people, including some actual couples, who took part in other studies.
But a heads up for prospective parents wanting a ripped Nobel laureate – the results were underwhelming.
Simulated embryos selected to be the smartest and tallest were only around 2.5 IQ points and 2.5 centimetres above average.
And if you don’t find mathematical embryos persuasive, the team checked their predictions in real people as well.
They had access to the genetic data and height stats of 28 unique families that had produced up to 20 kids apiece, with an average of 10.
Again, even if you wanted to create the mythical Übermensch, it seems you’re up against it.
In only seven of those families – just a quarter – did the person predicted to be tallest turn out to be so. In five, the person genetically ordained to be its tower of power was actually shorter than average.
“The notion that you could accurately choose your child’s height or select for a higher IQ, like in the movie Gattaca, has never been tested,” says Lencz.
“Through our research, we can confidently say that trait predictions for embryos based on polygenic scores are not very accurate.”
Why? Well, lots of reasons.
The base rate predictions from GWAS aren’t watertight in the first place. In even the largest studies those “polygenic scores” only explain about 5% of the variation in intelligence and 25% of the variation in height.
Then the nature versus nurture effect kicks in.
Parents with genetically higher intelligence are likely to create more learning opportunities in the home: books for example. That muddies the water on how much the IQ results in those GWAS are genetic or environmental.
Another issue, write the authors, is that subtle genetic changes called SNPs (single nucleotide polymorphisms) used to predict traits, are themselves influenced by the environment.
And then there is the issue that gene findings derived in one ethnic group may not faithfully apply to others.
Having said all that, gene studies are getting ever more comprehensive, vacuuming in more data – including sequencing of the whole genome – from more people.
The authors’ own modelling suggests, for example, that with access to data on 10 million folk, predictions could be refined to yield an expected IQ gain of around seven points.
But the aforementioned caveats still hold.
“There is much about these traits that is unpredictable,” says Shai Carmi, co-lead author on the study from Israel’s Hebrew University of Jerusalem.
“If someone selected an embryo that was predicted to have an IQ that was two points higher than the average, this is no guarantee it would actually result in that increase. There is a lot of variability that is not accounted for in the known gene variants.”
Meanwhile, the debate on whether genetic enhancement is simply eugenics rebranded will roll on.
But this study raises a here-and-now pointy issue for the informed consent of parents wanting to access these tests, which have been called “23andMe on embryos” (after the home genetic testing kit).
Those parents may want to ask their test provider: “what traits can we really predict with genetic testing?”
And listen carefully to the answer.