“Recent GWASs on other complex traits, such as height, body mass index, and schizophrenia, demonstrated that with greater sample sizes, the SNP h2 increases. [...] we suspect that with greater sample sizes and better imputation and coverage of the common and rare allele spectrum, over time, SNP heritability in ASB [antisocial behavior] could approach the family based estimates.”
I don’t know why Tielbeek says that, unless he’s confusing SNP heritability with PGS: a SNP heritability estimate is unconnected to sample size. Increasing n will reduce the standard error but assuming you don’t have a pathological case like GCTA computations diverging to a boundary of 0, it should not on average either increase or decrease the estimate… Better imputation and/or sequencing more will definitely yield a new, different, larger SNP heritability, but I am really doubtful that it will reach the family-based estimates: using pedigrees in GREML-KIN doesn’t reach the family-based Neuroticism estimate, for example, even though it gets IQ close to the IQ lower bound.
For example, the meta-analysis by Polderman et al. (2015, Table 2) suggests that 93% of all studies on specific personality disorders “are consistent with a model where trait resemblance is solely due to additive genetic variation”. (Of note, for “social values” this fraction is still 63%).
Twin analysis can’t distinguish between rare and common variants, AFAIK.
The SNP heritabilities I’m referring to are https://en.wikipedia.org/w/index.php?title=Genome-wide_complex_trait_analysis&oldid=871623331#Psychological There’s quite low heritabilities across the board, and https://www.biorxiv.org/content/10.1101/106203v2 shows that the family-specific rare variants (which are still additive, just rare) are almost twice as large as the common variants. A common SNP heritability of 10% is still a serious limit, as it upper bounds the PGS which will be available anytime soon, and also hints at very small average effects making it even harder. Actually, 10% is much worse than it seems even if you compare to the quoted IQ’s 30%, because personality is easy to measure compared to IQ, and the UKBB has better personality inventories than IQ measures (at least, substantially higher test-retest reliabilities IIRC).
Dominance...And what about epistasis? Is it just that there are quadrillions of possible combinations of interactions and so you would need astronomical sample sizes to achieve sufficient statistical power after correcting for multiple comparisons?
Yes. It is difficult to foresee any path towards cracking a reasonable amount of the epistasis, unless you have faith in neural net magic starting to work when you have millions or tens of millions of genomes, or something. So for the next decade, I’d predict, you can write off any hopes of exploiting epistasis to a degree remotely like we already can additivity. (Epistasis does make it a little harder to plan interventions: do you wind up in local optima? Does the intervention fall apart in the next generation after recombination? etc. But this is minor by comparison to the problem that no one knows what the epistasis is.) I’m less familiar with how well dominance can work.
----
So to summarize: the SNP heritabilities are all strikingly low, often <10%, and pretty much always <20%. These are real estimates and not anomalies driven by sampling error, nor largely deflated by measurement error. The PGSes, accordingly, are often near-zero and have no hits. The affordable increases in sample sizes using common SNP genotyping will push it up to the SNP heritability limit, hopefully; but for perspective, recall that IQ PGSes 2 years ago were *already* up to 11% (Allegrini et al 2018) and still have at least 20% to go, and IQ isn’t even that big a GWAS success story (eg height is >40%). The ‘huge success’ story for personality research is that with another few million samples years and years from now, they can reach where a modestly successful trait was years ago before they hit a hard deadend and will need much more expensive sequencing technology in generally brandnew datasets, at which point the statistical power issues become far more daunting (because rare variants by definition are rare), and other sources of predictive power like epistatic variants will remain inaccessible (barring considerable luck in someone coming up with a method which can actually handle epistasis etc). The value of the possible selection for the foreseeable future will be very small, and is already exceeded by selection on many other traits, which will continue to progress more rapidly, increasing the delta, and making selection on personality traits an ever harder sell to parents since it will largely come at the expense of larger gains on other traits.
Could you select for personality traits? A little bit, yeah. But it’s not going to work well compared to things selection does work well for, and it will continue not working well for a long time.
I don’t know why Tielbeek says that, unless he’s confusing SNP heritability with PGS: a SNP heritability estimate is unconnected to sample size. Increasing n will reduce the standard error but assuming you don’t have a pathological case like GCTA computations diverging to a boundary of 0, it should not on average either increase or decrease the estimate… Better imputation and/or sequencing more will definitely yield a new, different, larger SNP heritability, but I am really doubtful that it will reach the family-based estimates: using pedigrees in GREML-KIN doesn’t reach the family-based Neuroticism estimate, for example, even though it gets IQ close to the IQ lower bound.
Thanks, all of that makes sense, agree. I also wondered why SNP heritability estimates should increase with sample size.
To summarize, my sense is the following: Polygenic scores for personality traits will likely increase in the medium future, but are very unlikely to ever predict more than, say, ~25% of variance (and for agreeableness maybe never more than ~15% of variance). Still, there is a non-trivial probability (>15%) that we will be able to predict at least 10% of variance in agreeableness based on DNA alone within 20 years, and more than >50% probability that we can predict at least 5% of variance in agreeableness within 20 years from DNA alone.
Or do you think these predictions are still too optimistic?
But couldn’t one still make use of rare variants, especially in genome synthesis? Maybe also in other settings?
The value of the possible selection for the foreseeable future will be very small, and is already exceeded by selection on many other traits, which will continue to progress more rapidly, increasing the delta, and making selection on personality traits an ever harder sell to parents since it will largely come at the expense of larger gains on other traits.
I agree that selecting for IQ will be much easier and more valuable than selecting for personality traits. It could easily be the case that most parents will never select for any personality traits.
However, especially if we consider IES or genome synthesis, even small reductions in dark personality traits—such as extreme sadism—could be very valuable from a long-termist perspective.
For example, assume it’s 2050, IES is feasible and we can predict 5% of the variance in dark traits like psychopathy and sadism based on DNA alone. There are two IES projects: IES project A only selects for IQ (and other obvious traits relating to e.g. health), IES project B selects for IQ and against dark traits, otherwise the two projects are identical. Both projects use 1-in-10 selection, for 10 in vitro generations.
According to my understanding, the resulting average psychopathy and sadism scores of the humans created by project B could be about one SD* lower compared to project A. Granted, the IQ scores would also be lower, but probably by no more than 2 standard deviations (? I don’t know how to calculate this at all, could also be more).
It depends on various normative and empirical views whether this is worth it, but it very well might be: 180+IQ humans with extreme psychopathy or sadism scores might substantially increase all sorts of existential risks, and project A would create almost 17 times** as many such humans compared to project B, all else being equal.
The case for trying to reduce dark traits in humans created via genome synthesis seems even stronger.
One could draw an analogy with AI alignment efforts: Project A has a 2% chance of creating an unaligned AI (2% being the prevalence of humans with psychopathy scores 2 SDs above the norm). Project B has only a 0.1% chance of creating an unaligned AI. Project B is often preferable even if it’s more expensive and/or its AI is less powerful.
*See the calculation in my above comment: A PGS explaining 4% of variance in a trait can reduce this trait by 0.2 standard deviations in one generation. This might enable 1 SD (?) in 10 in vitro generations; though I don’t know, maybe one would run out of additive variance long before?
**pnorm(12, mean=10, sd=1, lower.tail=FALSE) / pnorm(12, mean=9, sd=1, lower.tail=FALSE) = 16.85. This defines extreme psychopathy and/or sadism as being 2 SDs or more above the norm, assumes that these traits are normally distributed, and that project B indeed has average scores of 1SD less than project A. (It also assumes IQ means for the two projects are identical, which is not realistic.)
I don’t know why Tielbeek says that, unless he’s confusing SNP heritability with PGS: a SNP heritability estimate is unconnected to sample size. Increasing n will reduce the standard error but assuming you don’t have a pathological case like GCTA computations diverging to a boundary of 0, it should not on average either increase or decrease the estimate… Better imputation and/or sequencing more will definitely yield a new, different, larger SNP heritability, but I am really doubtful that it will reach the family-based estimates: using pedigrees in GREML-KIN doesn’t reach the family-based Neuroticism estimate, for example, even though it gets IQ close to the IQ lower bound.
Twin analysis can’t distinguish between rare and common variants, AFAIK.
The SNP heritabilities I’m referring to are https://en.wikipedia.org/w/index.php?title=Genome-wide_complex_trait_analysis&oldid=871623331#Psychological There’s quite low heritabilities across the board, and https://www.biorxiv.org/content/10.1101/106203v2 shows that the family-specific rare variants (which are still additive, just rare) are almost twice as large as the common variants. A common SNP heritability of 10% is still a serious limit, as it upper bounds the PGS which will be available anytime soon, and also hints at very small average effects making it even harder. Actually, 10% is much worse than it seems even if you compare to the quoted IQ’s 30%, because personality is easy to measure compared to IQ, and the UKBB has better personality inventories than IQ measures (at least, substantially higher test-retest reliabilities IIRC).
Yes. It is difficult to foresee any path towards cracking a reasonable amount of the epistasis, unless you have faith in neural net magic starting to work when you have millions or tens of millions of genomes, or something. So for the next decade, I’d predict, you can write off any hopes of exploiting epistasis to a degree remotely like we already can additivity. (Epistasis does make it a little harder to plan interventions: do you wind up in local optima? Does the intervention fall apart in the next generation after recombination? etc. But this is minor by comparison to the problem that no one knows what the epistasis is.) I’m less familiar with how well dominance can work.
----
So to summarize: the SNP heritabilities are all strikingly low, often <10%, and pretty much always <20%. These are real estimates and not anomalies driven by sampling error, nor largely deflated by measurement error. The PGSes, accordingly, are often near-zero and have no hits. The affordable increases in sample sizes using common SNP genotyping will push it up to the SNP heritability limit, hopefully; but for perspective, recall that IQ PGSes 2 years ago were *already* up to 11% (Allegrini et al 2018) and still have at least 20% to go, and IQ isn’t even that big a GWAS success story (eg height is >40%). The ‘huge success’ story for personality research is that with another few million samples years and years from now, they can reach where a modestly successful trait was years ago before they hit a hard deadend and will need much more expensive sequencing technology in generally brandnew datasets, at which point the statistical power issues become far more daunting (because rare variants by definition are rare), and other sources of predictive power like epistatic variants will remain inaccessible (barring considerable luck in someone coming up with a method which can actually handle epistasis etc). The value of the possible selection for the foreseeable future will be very small, and is already exceeded by selection on many other traits, which will continue to progress more rapidly, increasing the delta, and making selection on personality traits an ever harder sell to parents since it will largely come at the expense of larger gains on other traits.
Could you select for personality traits? A little bit, yeah. But it’s not going to work well compared to things selection does work well for, and it will continue not working well for a long time.
Thanks, all of that makes sense, agree. I also wondered why SNP heritability estimates should increase with sample size.
To summarize, my sense is the following: Polygenic scores for personality traits will likely increase in the medium future, but are very unlikely to ever predict more than, say, ~25% of variance (and for agreeableness maybe never more than ~15% of variance). Still, there is a non-trivial probability (>15%) that we will be able to predict at least 10% of variance in agreeableness based on DNA alone within 20 years, and more than >50% probability that we can predict at least 5% of variance in agreeableness within 20 years from DNA alone.
Or do you think these predictions are still too optimistic?
Interesting, thanks.
But couldn’t one still make use of rare variants, especially in genome synthesis? Maybe also in other settings?
I agree that selecting for IQ will be much easier and more valuable than selecting for personality traits. It could easily be the case that most parents will never select for any personality traits.
However, especially if we consider IES or genome synthesis, even small reductions in dark personality traits—such as extreme sadism—could be very valuable from a long-termist perspective.
For example, assume it’s 2050, IES is feasible and we can predict 5% of the variance in dark traits like psychopathy and sadism based on DNA alone. There are two IES projects: IES project A only selects for IQ (and other obvious traits relating to e.g. health), IES project B selects for IQ and against dark traits, otherwise the two projects are identical. Both projects use 1-in-10 selection, for 10 in vitro generations.
According to my understanding, the resulting average psychopathy and sadism scores of the humans created by project B could be about one SD* lower compared to project A. Granted, the IQ scores would also be lower, but probably by no more than 2 standard deviations (? I don’t know how to calculate this at all, could also be more).
It depends on various normative and empirical views whether this is worth it, but it very well might be: 180+IQ humans with extreme psychopathy or sadism scores might substantially increase all sorts of existential risks, and project A would create almost 17 times** as many such humans compared to project B, all else being equal.
The case for trying to reduce dark traits in humans created via genome synthesis seems even stronger.
One could draw an analogy with AI alignment efforts: Project A has a 2% chance of creating an unaligned AI (2% being the prevalence of humans with psychopathy scores 2 SDs above the norm). Project B has only a 0.1% chance of creating an unaligned AI. Project B is often preferable even if it’s more expensive and/or its AI is less powerful.
*See the calculation in my above comment: A PGS explaining 4% of variance in a trait can reduce this trait by 0.2 standard deviations in one generation. This might enable 1 SD (?) in 10 in vitro generations; though I don’t know, maybe one would run out of additive variance long before?
**pnorm(12, mean=10, sd=1, lower.tail=FALSE) / pnorm(12, mean=9, sd=1, lower.tail=FALSE) = 16.85. This defines extreme psychopathy and/or sadism as being 2 SDs or more above the norm, assumes that these traits are normally distributed, and that project B indeed has average scores of 1SD less than project A. (It also assumes IQ means for the two projects are identical, which is not realistic.)