3 ms·

The obligatory article on this subject is, as always: https://www.gwern.net/Embryo-selection https://www.gwern.net/Embryo-selection It's long, but there's a c

The obligatory article on this subject is, as always: https://www.gwern.net/Embryo-selection https://www.gwern.net/Embryo-selection It's long, but there's a concise summary at the top that's definitely worth a read. Embryo selection works best when there are a very small number of genetic variants that you want to select over. Want a child to not be born with Tay-Sachs disease? Can do! Want a child to be smarter or better-looking? That's going to be a lot harder, since those traits are influenced by a huge number of genetic variants, each with small effect, and we've only got a handful of embryos to choose between, which limits the amount of influence we can have here.

Accujack 7y ago
Also, we honestly don't understand what "smarter" or "better looking" mean in genetic terms... we don't really even agree on what they mean between individuals. So, doing embryo culling works for specific, simple, detectable conditions, but the idea that it could be anything else without a lot of knowledge we as a species don't have is delusional at this point in time.
- klipt 7y ago
  > we don't really even agree on what they mean between individuals. Individual people can always disagree, but average opinion can be quantified. Take a bunch of pictures, throw them on Tinder and see how many matches they get. Although for men height is often a bigger factor than small variations in looks. Allow embryo selection for 6'+ sons, that's where the real money will be!
  - jakobegger 7y ago
    > Allow embryo selection for 6'+ sons, that's where the real money will be! Yes, but height is influenced by hundreds of genes, each with a minuscule influence, which will make it all but impossible to create a model that predicts height accurately. Additionally, since all the embryos you can select from are from the same two parents, models created from population samples are going to be a lot less useful. So I guess you won't be able to select for tall sons, at best you might be able to pick the larger of two embryos with slightly higher than random chance.
    - pjscott 7y ago
      All but impossible? We got a model that predicts height startlingly well, usually to within a few centimeters, back in 2017: https://infoproc.blogspot.com/2017/09/accurate-genomic-prediction-of-human.html https://infoproc.blogspot.com/2017/09/accurate-genomic-predi...
    - gwern 7y ago
      > Yes, but height is influenced by hundreds of genes, each with a minuscule influence, which will make it all but impossible to create a model that predicts height accurately. Not at all. The current PGSes for height already predict ~40% variance (or r=0.63): see https://www.biorxiv.org/content/early/2017/09/19/190124 https://www.biorxiv.org/content/early/2017/09/19/190124 or https://www.biorxiv.org/content/early/2018/07/24/375337 https://www.biorxiv.org/content/early/2018/07/24/375337 > Additionally, since all the embryos you can select from are from the same two parents, models created from population samples are going to be a lot less useful. No? It's all genes. The genes within a family are the same as across the population. The PGS is reduced by half* simply because you are comparing half-related embryos (siblings), so they will share many of the same genes and there are no differences to select on, but that's the nature of it. You could do selection on unrelated embryos if you really wanted to, but few people seem to want to... * between-sibling PGS reduction is sometimes more depending on other factors like 'nature of nurture'; this is something that affects EDU more than IQ, but other traits like BMI or height appear to not be affected > So I guess you won't be able to select for tall sons, at best you might be able to pick the larger of two embryos with slightly higher than random chance. It'd actually be 65% with the current height PGSes (upperbounded at perhaps 71% as WGS can recover the full heritability: https://www.biorxiv.org/content/10.1101/588020v1 https://www.biorxiv.org/content/10.1101/588020v1 ).
      - DuskStar 7y ago
        > > So I guess you won't be able to select for tall sons, at best you might be able to pick the larger of two embryos with slightly higher than random chance. > It'd actually be 65% with the current height PGSes (upperbounded at perhaps 71% as WGS can recover the full heritability: https://www.biorxiv.org/content/10.1101/588020v1 https://www.biorxiv.org/content/10.1101/588020v1 ). Wouldn't knowing 65% of the variance let you pick the best of two more than 75% of the time? (my intuition would be 50 + 65/2 %, but I'm probably wrong there)
toasterlovin 7y ago
FWIW, your last paragraph is a little misleading. I think a better way to state it is that you can be 100% certain of selecting embryos that do not have diseases caused by specific, known single genes. Selection can still be extremely effective on polygenic traits (traits influenced by a huge number of different genes), it's just that there isn't a single switch that you're turning on or off. After all, you're selecting the most intelligent (or, more specifically, the most likely to be the most intelligent) out of several dozen embryos. That is like artificial selection on steroids. And I should add: in theory. Right now we still don't know all the various genes that are responsible for intelligence and, as I understand it, there's a good chance that a significant number of the genes involved in a polygenic trait like intelligence are what are called rare variants. IE, they're not widely present in the population and are instead specific to certain lineages. If that's the case, then it will be hard to identify those and select the ones present in you and your mate's respective lineages.
- pjscott 7y ago
  The article I linked put it way better than I did, so I'll just quote that for clarification: > One of the key problems is that polygenic scores are the sum of many individual small genes’ effects and form a normal distribution, which is tightly distributed around a mean; the further into the tail one goes, the larger a sample it requires to realize a gain—to put it another way, if you have 10 samples, it’s easy (a 1 in 10 probability) that your next random sample will be the largest sample yet, but if you have 100 samples, now the probability of an improvement is the much harder 1 in 100, and if you have 1000, it’s only 1 in 1000, and worse, if you luck out and there’s an improvement, the improvement is ever tinier.
- jakobegger 7y ago
  > That is like artificial selection on steroids No, it's not, because you are selecting on genotype rather than phenotype. Embryo selection only works when there is a simple connection between genotype and phenotype (eg. what type of earlobes do you prefer). For complex traits like "intelligence", selecting based on genotype is not going to work. To much noise, too little signal.

Accujack 7y ago

Also, we honestly don't understand what "smarter" or "better looking" mean in genetic terms... we don't really even agree on what they mean between individuals. So, doing embryo culling works for specific, simple, detectable conditions, but the idea that it could be anything else without a lot of knowledge we as a species don't have is delusional at this point in time.

klipt 7y ago
> we don't really even agree on what they mean between individuals. Individual people can always disagree, but average opinion can be quantified. Take a bunch of pictures, throw them on Tinder and see how many matches they get. Although for men height is often a bigger factor than small variations in looks. Allow embryo selection for 6'+ sons, that's where the real money will be!
- jakobegger 7y ago
  > Allow embryo selection for 6'+ sons, that's where the real money will be! Yes, but height is influenced by hundreds of genes, each with a minuscule influence, which will make it all but impossible to create a model that predicts height accurately. Additionally, since all the embryos you can select from are from the same two parents, models created from population samples are going to be a lot less useful. So I guess you won't be able to select for tall sons, at best you might be able to pick the larger of two embryos with slightly higher than random chance.
  - pjscott 7y ago
    All but impossible? We got a model that predicts height startlingly well, usually to within a few centimeters, back in 2017: https://infoproc.blogspot.com/2017/09/accurate-genomic-prediction-of-human.html https://infoproc.blogspot.com/2017/09/accurate-genomic-predi...
  - gwern 7y ago
    > Yes, but height is influenced by hundreds of genes, each with a minuscule influence, which will make it all but impossible to create a model that predicts height accurately. Not at all. The current PGSes for height already predict ~40% variance (or r=0.63): see https://www.biorxiv.org/content/early/2017/09/19/190124 https://www.biorxiv.org/content/early/2017/09/19/190124 or https://www.biorxiv.org/content/early/2018/07/24/375337 https://www.biorxiv.org/content/early/2018/07/24/375337 > Additionally, since all the embryos you can select from are from the same two parents, models created from population samples are going to be a lot less useful. No? It's all genes. The genes within a family are the same as across the population. The PGS is reduced by half* simply because you are comparing half-related embryos (siblings), so they will share many of the same genes and there are no differences to select on, but that's the nature of it. You could do selection on unrelated embryos if you really wanted to, but few people seem to want to... * between-sibling PGS reduction is sometimes more depending on other factors like 'nature of nurture'; this is something that affects EDU more than IQ, but other traits like BMI or height appear to not be affected > So I guess you won't be able to select for tall sons, at best you might be able to pick the larger of two embryos with slightly higher than random chance. It'd actually be 65% with the current height PGSes (upperbounded at perhaps 71% as WGS can recover the full heritability: https://www.biorxiv.org/content/10.1101/588020v1 https://www.biorxiv.org/content/10.1101/588020v1 ).
    - DuskStar 7y ago
      > > So I guess you won't be able to select for tall sons, at best you might be able to pick the larger of two embryos with slightly higher than random chance. > It'd actually be 65% with the current height PGSes (upperbounded at perhaps 71% as WGS can recover the full heritability: https://www.biorxiv.org/content/10.1101/588020v1 https://www.biorxiv.org/content/10.1101/588020v1 ). Wouldn't knowing 65% of the variance let you pick the best of two more than 75% of the time? (my intuition would be 50 + 65/2 %, but I'm probably wrong there)

toasterlovin 7y ago

FWIW, your last paragraph is a little misleading. I think a better way to state it is that you can be 100% certain of selecting embryos that do not have diseases caused by specific, known single genes. Selection can still be extremely effective on polygenic traits (traits influenced by a huge number of different genes), it's just that there isn't a single switch that you're turning on or off. After all, you're selecting the most intelligent (or, more specifically, the most likely to be the most intelligent) out of several dozen embryos. That is like artificial selection on steroids. And I should add: in theory. Right now we still don't know all the various genes that are responsible for intelligence and, as I understand it, there's a good chance that a significant number of the genes involved in a polygenic trait like intelligence are what are called rare variants. IE, they're not widely present in the population and are instead specific to certain lineages. If that's the case, then it will be hard to identify those and select the ones present in you and your mate's respective lineages.

pjscott 7y ago
The article I linked put it way better than I did, so I'll just quote that for clarification: > One of the key problems is that polygenic scores are the sum of many individual small genes’ effects and form a normal distribution, which is tightly distributed around a mean; the further into the tail one goes, the larger a sample it requires to realize a gain—to put it another way, if you have 10 samples, it’s easy (a 1 in 10 probability) that your next random sample will be the largest sample yet, but if you have 100 samples, now the probability of an improvement is the much harder 1 in 100, and if you have 1000, it’s only 1 in 1000, and worse, if you luck out and there’s an improvement, the improvement is ever tinier.
jakobegger 7y ago
> That is like artificial selection on steroids No, it's not, because you are selecting on genotype rather than phenotype. Embryo selection only works when there is a simple connection between genotype and phenotype (eg. what type of earlobes do you prefer). For complex traits like "intelligence", selecting based on genotype is not going to work. To much noise, too little signal.