I’m a software developer by training with an interest in genetics. I currently run a startup working on multiplex gene editing technology.
GeneSmith
Karma: 6,125
Do any of the companies expose the variance on individual predictions or only the point estimates? 9% ± 0.5% is a different decision than 9% ± 4% when the alternative embryo is at 7%, and i’d guess most parents would rather take a lower mean embryo with tight variance than a higher-mean one with wide variance if the report actually let them see that, if the variance is in the underlying predictor but not in the report, that seems like a cheap product fix imo
Ok, I’ve literally checked with the most knowledgeable people in the world on this, and the answer is that prediction error (i.e. the variance in outcomes), does vary a tiny but, but only as a result of differences in the imputation accuracy of each embryo’s genome.
Herasight is the only company that does embryo genome imputation at the moment, so this isn’t even relevant to Genomic Prediction or Orchid.
Also, thinking about “error” in the prediction only really makes sense for continuous traits like height or IQ. For disease risk, any company worth dealing with account for prediction error in their risk prediction.
A weaker predictor should always produce a disease risk that’s close to the population average for that particular embryo’s sex and genetic ancestry. A stronger one will show more variance in risk from one embryo to the next.
And this is in fact exactly what you see. There’s quite a large spread in the risk of a disease like type 1 diabetes, and quite a small spread for something like epithelial ovarian cancer.
For IQ, the standard deviation of IQ for a randomly chosen embryo would be +-10.6 (15/sqrt(2)). If you know the predicted IQ from Herasight’s IQ predictor, that variance shrinks to 9.3 because a portion of the variance is now “fixed” thanks to the prediction.
For example, an embryo with a predicted IQ of +10 would have a narrowed, right shifted distribution compared to an average embryo: https://pbs.twimg.com/media/G6oTYmLbwAI51py?format=jpg&name=medium
And yes, Herasight shows this. Neither Genomic Prediction nor Orchid offer selection on continuous traits right now. I don’t know if Nucleus does, but they have other issues such that confidence intervals should probably be the least of your concerns.
This is a fair critique. I think I’ve partially ended up training myself to respond too dismissively by spending a lot of time engaging with the attention dynamics of Twitter.
It may be technically possible to do PGT-P in Australia right now, but it’s very hard and I know several people whose data is basically being held hostage by the PGT company.
MAYBE you could do it if you could convince them to share the raw PGTA data with you and transfer an embryo of your choosing, but unless you have written pre-approval I wouldn’t go that route. Australia can ban you from sending your own embryos abroad.
I would recommend Australian patients fly abroad if they want to do PGT-P.
I’d love to get this changed. If you’ve got any contacts in Australia I or someone else in the industry could talk to I’d be happy to chat with them. Even if they’re not going to allow IQ, it’s pretty stupid to ban embryo selection against Alzheimer’s or schizophrenia when there’s no effective treatment for either.
What is the tradeoff between having a baby now and waiting a few years? Like if I have a baby know with the current technology, how many IQ points, risk of Alzheimer or depression etc. will the baby miss out on if it was instead born in five years?
Right now you can get about +8 points of IQ from 10 embryos. A theoretically perfect predictor could get you about +13.
For Alzheimer’s it’s probably a similar gap between current and theoretically optimal.
However I don’t think we’re going to hit the ceiling of max gains in the next few years. We need way more data for that and no one is on track to gather that much.
My guess is IQ gain might get up to 9 or 10 in the next couple of years, and we’ll continue to see marginal improvements in other stuff like diseases.
The tech for adults is much harder. We continue to see improvements in therapeutics for a lot of things like high cholesterol, cancers, etc. So I’m hopeful those will actually make many of the diseases we can currently screen for much more treatable in a few decades.
As for doing genetic modification in adults, we’ve got the most rudimentary possible stuff right now. Probably the most interesting one from a layperson point of view is Verve-102, which knocks out PCSK9 in the liver, permanently lowering cholesterol. I think this would likely have positive impacts for almost anyone who got it, though the largest benefits will accrue to people with a family history of high cholesterol.
For more systemic, polygenic stuff, the best bet is probably something like what R3 Bio is doing growing “organ sacks”. If you can grow new organs (or potentially even a new body without a brain), you could genetically engineer it to be resistant to almost any disease you can think of.
I want to write more about this before too long.
“Importantly, there is limited evidence supporting almost all the claims, however well intentioned. That PGT-P improves real-world child health outcomes in a way that justifies its routine use is unproven”
I hear this claim repeated often by people in the field, most of whom seem unaware that there are multiple papers validating polygenic scores in a within-family context. Just to list a few:
Lello et al Wolfram et al Moore et al Plomin et al Selzam et al
Maybe you’re still thinking “that doesn’t prove these actually work in a clinical context”. But we already have polygenic predictors deployed in clinical practice, such as Myriad’s breast cancer predictor.
But maybe you’re still not convinced. Maybe you think “we need to wait for a bunch of selected embryos to grow up, then observe whether or not selection worked.” In that case, I’d just point out that this dataset already exists: it’s called “siblings”, and it can show you exactly how well predictors perform on selecting an embryo with a lower risk of breast cancer or a higher predicted IQ.
“where is the reliable measure of creativity, practical judgment, personality, motivation and all the other social skills?”
The biobanks from which these predictors are trained have not yet deemed it worthwhile to examine the genetics of these other traits. So for the time being, we’re limited to diseases, IQ, height, and (maybe soon) personality and possibly facial appearance (the latter is still speculative at this point).
The other things you pointed out are of course important too. Motivation is, I think, particularly important. We’ll likely be able to test for this weakly soon via conscienciousness. But there are just obviously many other important human traits that we don’t have good predictors for at all. I think it’s a shame.
But the only way to solve this is with more data, which is too expensive to collect for the moment.
And super babies? Ranking embryos? It feels like market-driven eugenic thinking. I thought we’d moved on from that early to mid-20th century aberration.
Was anyone signing up to be part of 20th century eugenics? Of course not. It wasn’t a voluntary process at all. It was state sponsored sterilization and murder.
If you don’t see the difference between that and embryo selection, you either haven’t thought deeply about it or you’re catholic.
If it’s the latter, I understand. One day we’ll have a way to do this without any excess embryos. But not yet.
The positive manifold between different cognitive tests (which is what IQ actually is) is one of the most widely replicated findings in psychometrics. It predicts everything from your grades in school, to your odds of getting into a graduate program, to your lifetime earnings, to your odds of getting divorced.
It’s more strongly correlated with some things vs others (for example it’s more predictive of school grades than it is of income), but it’s one of the single most important factors influencing the trajectory of your life.
Obviously there are many other important aspects of life, but IQ influences a lot of them!
I debated whether to use the term “superbabies”. It has been used by quite a few of my friends who are having polygenically screened children and it a little more legible than the term “polygenic screening”. So I ultimately opted to use it for that reason.
Maybe it would have been better to title this something like “How to Have Polygenically Screened Children: 2026 edition”.
I can’t speak for other PGT companies, but I know Herasight will do genetic testing on multiple prospective egg or sperm donors for clients.
I don’t know of a single egg or sperm bank that does this kind of in depth testing by default. They don’t even offer basic polygenic scores, let alone expanded carrier screening.
You can’t screen egg directly because there’s no way to read the genome without destroying it, and eggs only have one copy.
You could in theory reconstruct the donor’s genome if you had enough of her eggs, but that would require buying and fertilizing a bunch of them.
It’s easiest to just directly ask the donor to get genetically sequenced.
Has anyone been able to actually access Fertilo anywhere in the world?
Yes, I’ve talked to one of the doctors in Australia who ran their clinical trial. He’s had several patients that have used it.
One of the barriers to access right now is it’s only approved for use with low-stim cycles, or “Mini-IVF”. Those kind of cycles naturally produce fewer eggs, and more immature eggs. Fertilo should work identically for regular stim cycles, but I don’t think there have been any clinical trials for it yet.
Herasight does height, IQ, and a few other non-disease triats like eye color and hair color. Stuff like facial appearance or athletic performance aren’t well enough understood yet for anyone to offer precitors. They’ll probably have personality predictors at some point in the next year or so, but that’s TBD.
Nucleus offers stuff like this too, but for reasons explained in the article, I would avoid using them for the time being.
Orchid and Genomic Prediction only offer screening for disease related things at the moment.
Central asians are probably somewhere between east asians, middle easterners, and Europeans genetically, so I’d guess you’d see a reduction of maybe 15-25% in the expected gain for them relative to Europeans.
If I pick the former and the die roll goes badly, I know I’m directly responsible for that, for the rest of my life. If I pick the latter, then, every single time my kid misses the game-winning catch by a hair, or gets waitlisted to his dream school, I think about my paranoia being the reason his life is worse.
I don’t think this is really that much different from parents who blame themselves for their child not performing well on an exam because they didn’t sign them up for a private tutor. Embryo selection allows parents to influence the outcomes to some degree, but there’s still chance involved both from the remaining genetics we don’t understand and from the environment.
Humans didn’t evolve to live with that kind of damning empirical guilt.
I think humans evolved to deal with far worse than this. Think about how many children literally died from diseases, predation, and other horrible causes in the past. I think that stuff was way worse than seeing your child miss the game-winning field goal or even seeing them develop some mental disorder. And yet humans survived it.
Humans are strong!
The Practical Guide to Superbabies
This was possibly my favorite conference ever. Looking forward to the one this year.
The new custom iframe widgets are my favorite part of this update. I’ll be using them extensively as part of my upcoming post.
It’s true, I didn’t really address risks of egg freezing. I was focusing mostly on the benefits of freezing at a younger vs older age, but considering some women reading this will probably be freezing electively when they otherwise might have conceived naturally, this is probably worth addressing.
Egg freezing does come with some risks. The most notable is ovarian hyperstimulation syndrome, which happens when your tissues become extra permeable due to high levels of VGF, leading to fluid buildup in the abdomen.
It’s not fun. Most women experience some level of bloating and mood swings during egg freezing thanks to the hormone injections, but most of the time they’re just unpleasant.
Occasionally they can become much more severe. This especially happens in women whose estrogen goes to very high levels.
In rare cases (probably about 0.1-0.5%), hospitalization is necessary. There are steps you can take to reduce the odds of this, such as using a lupron only trigger, taking cabergoline after your retrieval, taking letrozole to keep your estrogen levels under control, and maybe one or two other things.
To be blunt, egg freezing IS a sacrifice. It’s one with an awesome outcome at the end of the tunnel: you get to have a baby, and if you do embryo selection you can even have a baby that’s smarter or less likely to get sick or less likely to develop a mental illness.
For some women this won’t be worth it. But considering the physical toll of pregnancy is almost certainly higher, I think unless you have a specific issues that makes egg freezing more dangerous for you (ex: some kind of estrogen-sensitive cancer), it’s worth at least considering if you can afford it and want kids.
Hmm… I’m not sure I agree with this. Open source Chinese models are pretty close to the performance ofChatGPT and Claude. And it seems to be pretty easy to distill a frontier model by training on its outputs.
I also think there is another risk from AI which is under discussed which is massive centralization of power enabled by AI. If the Yuskowskian model of AI danger proves incorrect and you can in fact control narrowly superhuman AI well enough to get it to do your alignment homework, the next biggest threat is that one small group of people control everything forever.
Are there any risks associated with the eggs being frozen for longer?
So far as we know, there are no additional risks. The main risk is you spend a lot of money on keeping them frozen for longer.
Can you freeze an egg for 20 years?
Almost certainly yes
Or even if there are some risks, is it the case that over any timespan you’d rather have an egg in the freezer than in the woman?
If you’re planning to have your final kid within a year, freezing eggs doesn’t really make sense unless you’re older and you need to bank a lot of eggs to have a good shot at having that kid.
For anything over a year, freezing eggs makes sense for fertility preservation.
We do not know and can not know what the long term effects “polygenic embryo selection”—or “unnatural selection” will be.
I think you could make the same argument about almost any technology. Was it foreseeable that the transistor would lead to the rise of cell phones and social media? Maybe to a small number of highly prescient people, but it certainly wasn’t obvious to most people.
However in the case of embryo selection I think the near term effects at least are pretty foreseeable because we can look at real siblings, see which ones have better polygenic scores for various things, and then check whether those people experience other serious downsides.
The answer is mostly “no”. Siblings with higher polygenic scores for IQ, for example, tend to have lower rates of mental illness, longer life expectancy, lower divorce rates, higher incomes, and lower rates of cardiovascular disease.
There are SOME exceptions to this; higher IQ correlates at about 0.15 with aspergers, and at about 0.06 with anorexia. But these exceptions are… exceptions. Most “good things” seem to run together at a genetic level.
None of these correlations are that strong, by the way. IQ only correlates with hypertension at about −0.07 or so.
And for diseases in general, you can see they overwhelmingly have positive correlations, meaning if you selected against one disease, chances are good that it will slightly decrease others.
A “better” approach: Accept that “Idiocracy” is a documentary film, ban IVF and let history take its course :-)
This seems like a bad approach to me.
Or if we want to fight “Idiocracy” naturally—make it a degree requirement, for PhD students, to have at least one child :-D
I’m guessing you’re just kind of trolling here, but i’ll answer like a true autist by taking this proposal seriously.
This is the bad kind of eugenics. We tried stuff like this in the 20th century and it overwhelmingly did not go well.
I worry about this too. One of my biggest concerns with embryo selection is that people in certain social circles will come to believe that you’re only a responsible parent if you have children via embryo selection, and that if they can’t they just won’t have them.
I think the best solution to this long term is to just make the tech really cheap and broadly available. But I also think we should try to encourage people to have more kids. Babies are great!
I made you a graph: https://imgur.com/a/v5EXnkf
You could probably do it for even less overseas