We can now edit the human genome – how far should we go? [video]
pbs.org
pbs.org
Also, as someone working with CRISPR, it isn't quite as easy as it is made to sound. It is relatively easy, but still a lot of work! I guess that is a footnote when it makes the impossible, possible though. We need to keep having these ethics conversations though, so we can use this tool wisely.
For further reading I would recommend Jennifer Doudna's "A crack in creation". She is one of the founders of CRISPR as a gene editing tool, and her book is good reading on the topic.
Then we'll need CAD tools. Really good CAD tools.
CRISPR is too unreliable and has too many off target effects to move the needle on height. By the time you've made a hundred CRISPR edits to a single genome, you've shot the other chromosomes full of holes. It's only useful for taking out one or two genes, for someone who knows they carry a fatal genetic disease, but still wants to have kids. True designer babies will be done with genome synthesis.
I understand it's fun to speculate about this.
Correct. We can only assemble synthetic bacteria right now. Synthetic sperm is years away. Synthetic eggs are too complicated, I wonder if they'll ever be made. Synthetic embryos would be true sci-fi stuff, requiring heavy engineering and many breakthroughs.
Emotional statement: "nobody is planning on"
Human germline engineering inspires strong feelings. I'm sure that the advent of motorized vehicles would have inspired the same strong feelings in horses, if they could have understood what was about to happen to them. Nevertheless.
Genetic engineering for human treatments took a 20+ year hit because one guy died in a trial at UPenn https://en.wikipedia.org/wiki/Jesse_Gelsinger
After He Jenkiu, genetic engineering took another huge hit because an irresponsible person ignored all advice and did a crappy job at fixing susceptibility to HIV infection.
This is not a "engines vs. horses argument". This is a "faster than light AI supership vs. horses argument".
Maybe we could clean up the off-target mutations using that.
Edit: To clarify, I mean somehow inducing crossover in a petri-dish.
CRISPR has been used successfully to treat Sickle Cell Anemia in at least one patient. [1]
You're spot on though that there are many obstacles to overcome (such as off target effects). CRISPR-Cas is not the answer, but rather just one more tool in the gene editing toolbox, albeit a very powerful one.
https://www.npr.org/sections/health-shots/2020/06/23/8775436...
Supposing we can up the precision to, say, 4 9s - i.e. 99.99%, would that suffice? I don't know enough to say anything, but given the number of cells we typically deal with, the chance of a wrong application proliferating, however small, if it gains an advantage over the intended mutation, what consequences would it have?
Disclaimer: I'm curious about this space, but don't know much right now and am just learning up.
Something Cas9 does really well is find the sequence you program it to look for, and then bind super tightly to it. You can even change your Cas9 so it will no longer break the DNA. Once you do this, Cas9 cannot edit the DNA by itself, but you can add a second enzyme to it that is able to change a single base of DNA from one letter to another. In theory this means you could target a single nucleotide somewhere in the ~3 billion letters of the genome. We call this changed Cas9 "base editor". It is much more refined than the crude DNA breakage of Cas9, since base editor will initiate only one time of edit in the genome (a single letter flipped from A to G or C to T). But it will still have unintended edits! It is pretty trivial to test for these in whole genome sequencing, and it turns out you will get plenty of them.
Another problem, which might seem more important, but I think is a much easier fix in the medium term (~10 years). Is that base editor will actually target a region of about 5-6 bases around the intended target and will happily flip those bases too. So you are unlikely to get just your intended mutation. Some of your cells will also get other mutations in the same spot. The reason this is less concerning in the long term is that I think this can be fixed with optimisation of the base editor protein. The off target effects are much more difficult to prevent.
>the off target effects are much more difficult to prevent
Why is that the case? And are there any promising theories for how we might overcome that problem?
CRISPR has already been approved for some forms of gene therapy. I assume that's because, for people with crippling diseases, the benefits clearly outweigh the risk. But going forward, as we address less crippling diseases, how do you grok the known benefit/unknown costs of using CRISPR for a cure? This will be an even bigger problem once CRISPR is used for more superficial or cosmetic things. We need to create a decision model and quantify the risks as much as possible.
I have a feeling that countries will "compete" on taking risks with CRISPR. For example, China might see CRISPR as a way to increase its dominance on the world stage. If there's a risky way for CRISPR to increase intelligence, they might take the risk, because China has more to gain than America.
I am still quite new to CRISPR so I am unsure what people are doing to solve this problem. As for why, I would say that the problem is that, like most proteins, Cas9 just bounces around using brownian motion and there is always just some probability of it being near the wrong piece of DNA at the right time. There's not much that can be done about that! However, we can do something about Cas9's binding affinity to sequences that are similar, but not idetnical, to the target sequence. One could imagine Cas9 binding (with lower affinity) to a sequence of 20 bases that is incorrect on a couple of bases. Indeed, much of the off-target effects come from this. So if one could modify Cas9 to only bind to the exact sequence that should reduce off-target mutations.
Technically the issue may or may not be drug "additive" and inhibitable like "stop making this protein" but "start making this with your blood cells".
It seems to me, viewed a certain way, the human body is almost entirely driven by emergent phenomenon. I have serious doubts about our ability to edit the source code to achieve very straight-forward single side-effect modifications.
We may very well be limited to merely adjusting "errant" DNA sequences that cause or are implicated in disease or undesired attributes. Who would choose a short child? Or one with myopia? Or one with weak muscles? Or even one with the misshapen chest and facial orifices necessary to survive lower partial pressures of oxygen?
The long term effect of custom genetic modification actually seems like it would just be used to reduce the overall genetic diversity of humans on the whole, not to improve it in drastic ways.
In fact, there are economic incentives to create broken clones that require daily medical attention so you can turn them into slaves that work for medicine.
Based on human history, I am fairly sure the end of humanity is approaching. 100 to 150 years I'd guess. I don't mean we'll be gone, just that what we'll become will not be recognized as human by our ancestors.
The point being hypothetical bigotries of the long dead is an unmeasurable metric (oxymoronic) and pretty useless.
Cat girls
edit: ahh... I love the part of HN that can't enjoy even the lightest of humor. FWIW: I have a severe chronic/borderline terminal health condition that is near certain to leave me dead quite early, not to mention the hell it's brought upon my physical body. In 10 years or so (maybe even sooner) - it will probably be something that's editable out in vitro. Also... I'm young as it is.
Let me have fun with my dreams of cat girls. Maybe you'd end up enjoying them too. :)
PS: The original comment brought a smile, but I still downvoted. Keeping most jokes and obvious comments off HN is a major long term benefit.
Sometimes it's not about what's visible to your eye...
Though, I do agree, in general, it's best practice not to have too many laughs on HN.
Also... yeah, current state of things is a bit hype. I have a lot of hope though having spoken with people doing research at places like University of Tokyo. I'm sure whether ethically or not, China has probably made some great strides we're not 100% aware of.
Virus would love that, yes why no, lets remove the only thing that has keep all living organism on this planet for billions of years from extinction.
Some country will allow it as it gives them a leg up on others. And once you can do it in X country, the pressure will be strong for others to follow suit and allow it or fall behind.
This is excepting the case where it generates some near term negative effect that outweighs the gains to be had by it.
If we don't go far enough, someone else will. Whether Iran or China or North Korea, someone else will eventually start producing superintelligent children that will outshine all western scientists and engineers[1]. If we don't start doing it too, we will become irrelevant, prehistoric tribes that have no place in the modern world.
If we decide to allow genetic enchancements, though, we need to ensure the society won't split into classes. Some parents will probably want the freedom to choose if their child is enchanced or not, but the unenchanced children will have no chance at anything in life. The only way forward is to require genetic enchancement, no matter the opposition from religious or other groups.
I am still shaken by that conclusion, but it seems inevitable. Irrelevance, inequality or force. Choose one, or they will choose for you.
[1] assuming science allows us to do this eventually
We just don't understand so much of genomic function through evolutionary time from the simple fact that we cannot easily observe it, that making anything more than a SNP change is asking for major, unfixable problems at a population level down the line. If risk is (threat x vulnerability x consequence), we need to practice the utmost caution when that vulnerability period is the rest of our species existence!
> Cas9 cannot edit the DNA by itself, but you can add a second enzyme to it that is able to change a single base of DNA from one letter to another. In theory this means you could target a single nucleotide somewhere in the ~3 billion letters of the genome. ..A single letter flipped from A to G or C to T.
That seems like an answer to your question. The "resolution", or a single "unit", in a genome sequence is a nucleotide.
> But it will still have unintended edits!
As for the accuracy of CRISPR, two kinds of errors are mentioned. "Off-target" mutations that occur elsewhere in the sequence:
> It is pretty trivial to test for these in whole genome sequencing, and it turns out you will get plenty of them.
And mutations that occur in the same targeted location:
> ..[The] base editor will actually target a region of about 5-6 bases around the intended target and will happily flip those bases too.
¹ https://news.ycombinator.com/item?id=24459427
---
On further reading, a single "unit" for double-stranded D/RNA is the base pair:
> bp = base pair(s) — one bp corresponds to approximately 3.4 Å (340 pm) of length along the strand, and to roughly 618 or 643 daltons for DNA and RNA respectively.
(A picometer, "pm", is 1×10^−12 m, or one trillionth of a meter.)
@ImaCake does mention that CRIPSR is able to "target a single nucleotide", a single base of DNA.
For single-stranded D/RNA, the nucleotide is the unit, abbreviated nt (or knt, Mnt, Gnt).
We can't predict the outcome, even in out wildest dreams or worst nightmare, we'll always be wrong.
Are we afraid of uncertainties? Of course we are, evolution taught us to be.
Not doing everything we can would be in total contradiction with the behavior of humankind since the dawn of ages.
I am both an optimist and an atheist, there is no God and nothing is sacred.
The beauty of evolution is that this is a process without designers. There is simply no need for them.
Go far enough that parents sue companies because their child doesn't match the designer baby form they have filled out.
Go far enough that there will be racial violence between the "natural" and genetically engineered humans.
Too much ambition will bring severe consequences... Please don't go beyond what is absolutely necessary.
godtoshi's vision
Engineered vs not engineered could easily become the new class system where the rich-ish are able to produce “better” children and exponentially widen the gap.
Could also lose all semblance of uniqueness. Want your child to be a redhead with a 170 IQ? Okay, pay enough and it’s yours.
So I think the real question is not how far SHOULD we go, but instead how far WILL we go. And how quickly will we go there.