4 ms·

I'm pretty sure preimplantation diagnostics are in many cases done with next-gen sequencing these days. Also even accepting the idea that GWAS studies for IQ y

by entee 11y ago

I'm pretty sure preimplantation diagnostics are in many cases done with next-gen sequencing these days. Also even accepting the idea that GWAS studies for IQ yield real and useful results (a claim most people I know in the human genetics field find highly dubious) CRISPR isn't a panacea. I posted a more detailed description above, but in short CRISPR causes a massively disruptive amount of damage in a cell. It does it very accurately, but a double strand break is definitely a Big Deal for a cell. This means it only works when you've got lots of cells to play with, a near-impossible task with human embryology. CRISPR is an unbelievably useful tool for research and a huge advance, but it's not the Solution to Human Genetic Engineering. We're still a long ways from that.

gwern 11y ago
> Also even accepting the idea that GWAS studies for IQ yield real and useful results (a claim most people I know in the human genetics field find highly dubious) It's not 'highly dubious' at all, I don't know who is telling you that or why. The twin studies have told us for a century that there are a lot of intelligence genes for us to find; the family studies have told us that for a century; the GCTAs have told us for half a decade that there were a lot of SNPs for us to find; the polygenic scores from the early GWASes told us there were a lot of SNPs for us to find once sample sizes got big enough; when the sample sizes reached the hundreds of thousands, the hits started coming and many of them were on genes already associated with the brain; the hits replicated in other samples (often using different but still highly g-loaded constructs) and had predictive power within-family too; followup GWASes replicated them, and they've been shown in Swedes, Englishmen, Chinese, European-Americans, African-Americans; the phenome studies have shown that intelligence SNPs and polygenic scores predict not just other cognitive constructs (both phenotype and polygenic scores thereof) but things already correlated with phenotypic intelligence like longevity or schizophrenia; and so on. The results are about as dubious as the height GWAS results - not at all. > This means it only works when you've got lots of cells to play with, a near-impossible task with human embryology. We'll see what sort of yields we get; embryos also have the advantage of being small, smaller than a whole body or even an eye. I'm sure you already saw the Chinese paper... Since one gets a median of ~9 eggs from each egg extract and about half of them will be implantable, one could afford to lose a decent fraction of them in exchange for a big enough payoff, and check with sequencing.

by entee 11y ago

gwern 11y ago

> Also even accepting the idea that GWAS studies for IQ yield real and useful results (a claim most people I know in the human genetics field find highly dubious) It's not 'highly dubious' at all, I don't know who is telling you that or why. The twin studies have told us for a century that there are a lot of intelligence genes for us to find; the family studies have told us that for a century; the GCTAs have told us for half a decade that there were a lot of SNPs for us to find; the polygenic scores from the early GWASes told us there were a lot of SNPs for us to find once sample sizes got big enough; when the sample sizes reached the hundreds of thousands, the hits started coming and many of them were on genes already associated with the brain; the hits replicated in other samples (often using different but still highly g-loaded constructs) and had predictive power within-family too; followup GWASes replicated them, and they've been shown in Swedes, Englishmen, Chinese, European-Americans, African-Americans; the phenome studies have shown that intelligence SNPs and polygenic scores predict not just other cognitive constructs (both phenotype and polygenic scores thereof) but things already correlated with phenotypic intelligence like longevity or schizophrenia; and so on. The results are about as dubious as the height GWAS results - not at all. > This means it only works when you've got lots of cells to play with, a near-impossible task with human embryology. We'll see what sort of yields we get; embryos also have the advantage of being small, smaller than a whole body or even an eye. I'm sure you already saw the Chinese paper... Since one gets a median of ~9 eggs from each egg extract and about half of them will be implantable, one could afford to lose a decent fraction of them in exchange for a big enough payoff, and check with sequencing.