Berkeley Genomics Project @Reprogenetics_

Exciting talk from @PaternaBio. If they can mature sperm in the lab, men with severe infertility could have biological kids. The same tech could let us genetically screen sperm before conception, making embryo screening far more powerful.

Elan Bechor @elanbechor

a week ago

Very excited to be attending this talk from @PaternaBio founder @apastusz today!

0 0 11 7K 0

Kudos to @nucleusgenomics for sponsoring this IRB study testing whether screening embryos based on polygenic risk scores will decrease implantation rates compared to screening only on embryo morphology! x.com/TreffDNA/statu…

Nathan Treff @TreffDNA

2 weeks ago

Many IVF programs treat the ASRM Opinion as a guideline for polygenic embryo selection. I’m pleased to announce the launch of PRISM (Polygenic Risk-Informed Selection versus Morphology), one of the largest clinical studies of polygenic embryo analysis to date. More in thread.

2 1 10 956 1

Today we launch NeuroRisk, the first embryo test for autism and schizophrenia that looks for mutations in evolutionarily important genes. These mutations are individually rare but over 10% of people carry at least one mutation in this class.

@fabiusforemarch I'm certainly not aiming to start a company myself that does all that stuff in that article, if that's what you mean. I'm aiming to accelerate research in the related areas, which almost entirely means supporting other people's efforts.

New book: "Methods for strong human germline engineering" A few takeaways: 🧵 lesswrong.com/posts/2w6hjpta…

@fabiusforemarch Explain?

Happy Oct4 @velychko_s !

@slothtotheflame @thecaptain_nemo How feasible? I mean, get good sleep, eat reasonably, exercise reasonably, try caffeine and other things; what else?

@slothtotheflame @thecaptain_nemo That would be great, but I think not that huge and also not that much more feasible.

@thecaptain_nemo lesswrong.com/posts/jTiSWHKA… berkeleygenomics.org/Explore

@thecaptain_nemo Unfortunately, adult human intelligence amplification is both probably really technically difficult, e.g. because brain access is difficult; and also lacks a clear pathway (how do BCIs actually significantly increase insight?). Reprogenetics is workable but needs more resources.

@VesselOfSpirit x.com/BerkeleyGenomi…

Berkeley Genomics Project @Reprogenetics_

10 months ago

somewhere, in a nearby Everett Branch...

1 6 27 2K 4

> You cannot CAUSE IQ to increase when you don’t have the causal input. You can tweak correlating genes all you want and that by definition will not induce causation. But these polygenic scores are validating on sibling pairs. This is very strong evidence they are capturing causal affects. It's true that some of the predictive power comes from non-causal effects; it's just not that much of it. Mostly it is causal. (Generally, much of the obvious non-causal predictive power is subtracted away when constructing a PGS.)

> What the above 2 points mean is that you can keep selecting the embryo their system scores highest amongst 10 and it doesn’t mean they’ll be higher IQ. It’s even slimmer chance you’ll get two directionally correct in a row because like I mentioned there’s 80% influence that could go in the opposite direction theoretically This doesn't make much sense. Each time you do embryo selection, you increase the average expected IQ by a few points. It's true that there's a significant chance (say, 1/3 or 1/6) that the actual IQ will be a bit lower than the average would have been without embryo selection. But the change in expected IQ will still increase. (Probably sublinearly if you do it for 20 generations.)

@kanzure @maxhodak_ It would make a lot more sense for someone who's an expert to do that; in the past when I've written up proposals for specific projects where I'm not an expert it's generally a waste of effort, especially when there's no one intending to fund it. Intent to fund is very useful.

@KianSadeghi5 @maxhodak_ Uh, what is this? I don't think this counts as public. It seems to say I couldn't criticize the methods.

@KianSadeghi5 @maxhodak_ What about the IQ one?

@maxhodak_ Do you know anyone who might like to fund a third-party validator of commercial claims about polygenic embryo screening? Shouldn't be too expensive, but confidence in funding would help because most academics don't find it to be a very academically interesting project.

I agree that point 6 seems somewhat wrong; I didn't consider 5 and 6 much because they seem vague / undefined. I mean, in some senses it would be bad to "carve your progeny's IQ trajectory". There's a valid point in what OP said about the unexplained part; the more precise statement is that you can significantly affect the mean but not much affect the variance. Whether that counts as "control" or not I leave to you. But yeah, the 80% unexplained doesn't really bear that much on whether you can continue genomic vectoring over multiple generations, or more strongly in one generation.

@capitalismandf1 @YounisJoseph In order for reprogenetics to be trustworthy and trusted, the results should be explained clearly and accurately. There's a strong incentive to state results misleadingly to get more attention. It's good to push back against that.

Hi Kian, great to see more IQ PGS research! A few things: * You say "optimize embryo DNA for 14 IQ points". I think you're referring to the expected spread for 10 embryos, but this is a misleading way of stating these results. Expected gain is the logical number to share. (To be clear, I think that Herasight sometimes summarizes things in a similarly unhelpful way, which should also be called out.) Further, the relevant number is achievable births; 10 achievable births is quite optimistic (and/or expensive) for most people at the moment. * You say in the context of the IQ model that "The models were trained on 1,000,000+ people", but that contradicts the 500k number from the paper; you may be referring to the data for the height and BMI models, but that doesn't match the context. * You say "validated across 40,000+ siblings"; that's not for IQ though, right? * You say "Parents across the world now have the choice to substantially increase their child’s intelligence."---but there are quite substantial dropoffs for many ancestry groups for PGSes trained on mainly European-ancestry data. This is a great aspiration, but it overstates the current state of things. x.com/KianSadeghi5/s…

Kian Sadeghi @KianSadeghi5

a month ago

AI is rapidly getting smarter. Now, humanity can too. Today, @nucleusgenomics is announcing Vitruvian, our newest set of genetic optimization models. Vitruvian’s intelligence model can optimize embryo DNA for 14 IQ points — nearly a standard deviation. The models were trained

217 158 2K 434K 1K