Berkeley Genomics Project

Speakers for Reproductive Frontiers 2026 (June 16-18, Berkeley) include leaders in polygenic prediction, artificial placentas, in vitro gametogenesis, in vitro oocyte maturation, repro law, + more. Early bird tickets are $350 (until April 25). reproductivefrontiers.org

True. To pull the rope sideways regarding inequality, I would want to go into "How do we make it widely accessible?". Examples: * anti-trust restrictions * open science * licensing or platform service rather than siloing * a general culture of innovation aiming to make the tech cheaper and more effective * researching cross-ancestry PGSes * developing strong reprogenetics (which makes things less expensive #strong-gv-and-why-it-matters" target="_blank" rel="nofollow noopener">berkeleygenomics.org/articles/Metho…) * innovation-positive / not-too-burdensome regulation (I also think it's far far more likely to be highly democratic, as it would enable anyone who wants to, to have kids who are likely to be very cognitively capable.) (I haven't addressed your other points yet.)

These discussions do feel like walking through molasses--or rather, tugging back and forth on a rope that doesn't matter. It makes little moral sense to tell parents that they can't improve the lives of their future children because it would hurt someone else's feelings. However, if you think more about it, there are potential concrete harms; less people with some condition probably does, at least by default, mean less support for that condition (social acceptance, logistical accommodations, medical care, scientific research), at least a bit in expectation. But then again, still, curing one person's cancer has the same effect on other people with cancer, and you wouldn't tell someone not to cure their cancer for that reason! It makes no sense. And yet, the topic keeps getting brought up. Are the people who bring it up just utterly confused? An alternative explanation is that they are poorly articulating a legitimate concern, and improperly framing it vaguely as pushback against reprogenetics. That's why I prefer to ask the question "[Reprogenetics can fairly obviously be enormously beneficial; so] In what manner can we develop it so that it is very widely beneficial, and avoids harm as much as feasible, especially unethical harm?". For example, one underlying real worry people might have is that wanting to avoid your child having some condition might reflect a preexisting attitude in the parents, embolden that attitude in the parents, and affirm that attitude as a public social attitude. Example of such an attitude: "Being deaf means your life is less worthwhile / worthy.". Now, these days people usually don't come out and express that attitude, but that attitude actually was prevalent, functionally speaking, during the decades of 20th century eugenics, and was one of the top justifications for eugenic policies. So it's not just a paranoid fantasy. That leads to actual substantive questions, like "How do we avoid having or affirming those attitudes?". For example, it's crucial that the uncontrollability of human life (genetics, development, and environment) is very clear to parents. No one can come close to guaranteeing that your child won't have any of a wide variety of diseases or other conditions. And so you too have a deaf child, in a fractional probabilistic sense, even if you use reprogenetics; you too have a blind child; you too have an autistic child; etc. x.com/antonioregalad…

Do you see a good way to advance the important parts of the conversation without getting stuck chasing an impossibly wide consensus? Funders (gvt, philanthropy, VC) are worried about public views and regulation, and the area could use a lot more funding to make the technology safe and highly effective.

In regards to I. Glenn Cohen, of Harvard, he's a proponent of reprogenetic technology, but also beholden to the attitudes and priorities of academic bioethics. His "seat at the table" comment implies indefinite discussions with implausibly wide constituency before action.

Starting right now. Debate between I. Glenn Cohen (Harvard/reprogenetic bioethics) vs. Cathy Tie, 'serial' human germline editing startup entrepreneur. This should not be an fair fight, but Glenn Cohen is a nice guy...so thehastingscenter.org/callahanprogra…

@antonioregalado Embryo editing specifically, though, probably isn't that important for polygenic traits (where each individual variant has a small effect). x.com/BerkeleyGenomi…

The more dramatic application of germline editing, the one where you might get big gains, is frank enhancement and protection against complex diseases of older age, and this topic was not addressed in the debate.

@antonioregalado @brian_armstrong It's a hypothetical reprogenetics method, wherein you ensemble together chromosomes from several different source cells into one cell. It'd be surprisingly powerful. Happy to chat about it if you like. I've written about it here: #method-chromosome-selection" target="_blank" rel="nofollow noopener">berkeleygenomics.org/articles/Metho… berkeleygenomics.org/articles/Chrom…

@BerkeleyGenomic @brian_armstrong what is chromosome selection?

To people who could move large amounts of money: What would you have to see, to convince you to put millions of dollars into the reprogenetics space (embryo screening, in vitro gametogenesis, iterated stem cell editing, chromosome selection)? There are social and regulatory issues--but those probably can't be resolved if the science itself is lagging. The science could really use a ton more money and talent (and the talent needs the money). This could be a generationally beneficial set of technological breakthroughs, but it needs your help to become safe and highly effective. What information or authoritative validation would get you more interested?

Thanks to John for the talk! Thanks to @michaelcurzi for video editing. YouTube link: youtube.com/watch?v=307H56…

BGPodcast #2: John Quain is a bioethicist and formerly an advisor for Manhattan Genomics, which aimed to do research on embryo gene editing before it closed down. We discuss the ethics of embryo editing to prevent disease. YouTube link in reply. ⬇️ Timestamps: 0:00 Introduction 3:09 The Manhattan Project for Gene Editing 6:39 Germline vs. Somatic Editing 14:21 Patient Autonomy and Consent 25:06 Doubts and Risks: Mosaicism, Off-Target Edits, and Unknowns 33:03 Oversight, Regulation, and Review Bodies 42:15 Where to Draw the Line on Human Enhancement 52:12 Accessibility and Equity 1:05:19 Pushback on Germline Editing 1:19:50 Regulatory Strategy 1:30:08 Closing Thoughts

Like I said a couple times, it's an open question. See berkeleygenomics.org/articles/Some_… But yeah I think it would be quite strange if the bulk of these effects just suddenly stop working right around +2 SDs. I don't see how that makes sense, biologically. It makes sense that there would be lots of tiny improvements, lots of little tweaks that slightly affect many different systems, and that these don't mostly immediately overlap / interfere. Of course there would eventually be more and more strongly diminishing returns; if "400 IQ" even meant anything, I wouldn't think you can get there just by putting together all the IQ positive SNPs. But as one touchpoint, consider animal breeding. It just has really big effects really fast. Like, many many SDs of change (e.g. agricultural animal yield measured in various ways). IQ is a pretty different trait, but that's still a substantive point of comparison. Do you have any evidence that it would have sharply diminishing returns around +2 SDs on IQ?

That's the assumption that I'm questioning. One +2 person has alleles A,B,C. Another +2 person has C,D,E. You assume that if we select/edit an embryo to get A,B,C,D,E that will get you to be > +2. I'm not sure that's the case. I think maybe at +4 you need A,B,D,E,W,Z. We don't know that we need W,Z because we haven't sequenced enough people at > +4 to determine that combination. Put it another way, if we only do PGS for height using only Guatemalan men (160cm average), do you think we can identify enough SNPs that will make a man grow to be 200cm?

I think that, very roughly at even odds, given O($500 million) going to a couple dozen key reprogenetics science research projects, within something like 10 or 15 years, we could make it technically feasible for roughly any parents who want to, to have a child who is likely to grow up to be a Nobel-prize-winning intellect in one way or another. This may seem like an outlandish claim. The hedging is important; I'm not 90% confident of that. Maybe in vitro gametogenesis is hard and takes another 2 decades to do at high enough quality to safely make a baby; maybe chromosome selection is infeasible; maybe you can't figure out how to do 200 edits to stem cells in vitro while maintaining genomic integrity. Maybe the effects on IQ of genes are highly sublinear. But there's multiple somewhat-disjunctive biotech pathways to strong reprogenetics, and we already know more than enough about the genetics of IQ, assuming not highly sublinear effects as you genomically vector upward. See berkeleygenomics.org/articles/Visua… and berkeleygenomics.org/articles/Metho… for more detail. Happy to discuss / debate, DMs open. In a sane civilization, this stuff would be funded to the gills. The birthright of humanity is to grow up--which means becoming kinder, wiser, more loving, more conscious, more creative, more understanding, more sane, and also, yes, more intelligent. Only just now, in the 21st century, does humanity finally have almost all of the tools needed to give ourselves more brainpower. It's a shame to be dragging our feet.

@DavidLiaoCH Why do you think that the bulk of the hundreds of SNPs that slightly increase intelligence in the -2 to +2 range stop increasing intelligence above that range?

The hundreds of SNPs are collected from people who are at <= +2 SD. To find the true outliers at +4 or +5 you need to sequence those people. Nobody has dug up Von Neumann's body to find out what made him different. And I don't think Terence Tao has ever given a sample either. If your data set consists of <= +2 SDs, how would you know what do edit to reach +4 or +5? I think this is still a data problem.

We know hundreds of SNPs associated with intelligence. Editing in particular has major additional wrinkles (mainly that it's hard to do without causing more damage, and that you don't know which SNV is causal; also, generally you're potentially making non-wild-type haplotypes). But more powerful selection methods such as chromosome selection or iterated recombinant selection could avoid those problems while strongly selecting for the SNPs in question. This should work, though there plenty of unknowns such as diminishing returns / nonlinearity. We're looking at more like -2 to +2 SDs (in bulk) to identify these SNPs, and then extrapolating. It's easy to raise question marks about that extrapolation, and it's a legit open question, but I think the default hypothesis-class would be "these SNPs probably have lots of non-overlapping, non-interfering brain-benefiting effects and therefore we can go pretty far, whether or not it's actually near-linear".

@BerkeleyGenomic If the sperm and the egg both come from people who are at +2 SD, I think it's very possible to stay there and not regress to the mean. To do editing, you need to know what to edit. You really think we have enough samples of geniuses at +3 SD and above to identify those SNPs?

@lu_sichu Right, but I have a pretty high probability that it just works on the first go. The reason that's reasonable is that we have a ton of natural experiments (anyone alive, genotype + phenotype). You'd also need other elements to not be harmful to development (IVG, IVF).

@BerkeleyGenomic if you fail at it, it might take a long time to retry. like the feedback loop seems to need a lot of time?

@DavidLiaoCH Do you mean +2 SD on IQ using embryo selection? I highly doubt that. Even if you have IVG and make 1000 embryos, and you have a PGS with R=.6, you're still not there. I'm talking about more advanced technology: iterated meiotic selection, chromosome selection, iterated editing.

I think in 10 years we can reliably select for an embryo that will be at +2 SD. To go higher we would need a very large sample of existing people who are at +3, +4, and above. To my knowledge, there's no biobank that actively targeting this group. The Nobel Prize winner sperm bank is defunct. China is probably the only country that can collect this data easily. They already do extensive IQ testing to find gifted kids. They need to extend this effort to collect whole genome sequences.

I don't expect anyone to just plunk down $500 million or even $5 million. Many of the relevant projects don't exist and would take a bunch of organizing to set up, and there are probably several key talent gaps. What I do hope for, what I would ask of major funders, is to make some sort of legible statement like: > Over the next 5 years, I'd be willing to put at least $100 million towards worthy science projects that my biologist expert friend agrees would accelerate strong reprogenetics. That way scientists would know that this is the case, and view it as worthwhile to organize the relevant projects (because they might actually get funded); and junior scientists would know they could make a career in advanced reprotech / reprogenetics. This is why I want people to understand that there's a good chance we're not that far off from this technology working. That seems like something that's worth a try, in order to get out of gridlock around this research. x.com/BerkeleyGenomi…

@m_goes_distance Supporting the emerging field of reprogenetics, e.g. by making this: reproductivefrontiers.org DM'd you

I’m tired of being performative with this biotech thing so I’ll be direct we're opening Superhuman Fund II soon our biggest focus is human enhancement(human 2.0) we’re funding science that pushes the frontier of human capability. examples: - peptides & protocols for peak performance and longevity - rejuvenation tech that actually moves biological age - fertility/embryo tech that changes the genetic baseline - hardware + robotics that augment human output if you’re building in that direction, I wanna talk. and if you're investing, we're qualifying LPs. DM for deets. bio/acc.