Genetic repair via CRISPR can inadvertently introduce other defects | Hacker News Reader

Genetic repair via CRISPR can inadvertently introduce other defects

phys.org

111 pointsamichail65 comments

A lot of the graphics that explain how CRISPR work show this very careful snipping of a strand of DNA. Tiny little scissors making pinpoint accuracy revisions. I think that's lead to a deep misunderstanding for most people (including myself) on how this technology actually works.

CRISPR is amazing, but it's still a fairly blunt tool. The article talks about one way: guide RNA can often bind to sequences that are very similar to the intended sequence, but not exactly, meaning it's probable that for every intended cut you're making several unintended ones. The "scissor" action itself, Cas9 protein, is less like a scissor cut and more like a jagged tear, which can damage surrounding DNA.

The repair pathways themselves are also imperfect. It's not a copy/paste like infographics show, it's more like emergency duct taping broken ends together.

All of that stuff combined...Again, amazing technology, but I get extremely nervous when I hear people talk about introducing CRISPR-based gene editing into the human gene pool. The generational effects there are still entirely unpredictable at this stage, could be disastrous, and any actual paths to trying to roll things back would be deeply ethically fraught.

> the generational effects there are still entirely unpredictable at this stage

Something I've always wondered about CRISPR. It seems the legal case for introducing it is the freedom to perform "body hacking" on yourself. But, at least in this case, isn't it more like "generational hacking?" Maybe I am fundamentally misunderstanding the technology, but it's strange to think about what sort of protections should be put in place to prevent someone being born with severe defects due to messing with the genome.

I'm not sure about that...

Isn't restricting people from reproducing based on their genetic material known as Eugenics?

This is the "ethically fraught" bit I was referencing. Once this genie is out of the bottle and into the gene pool, there's effectively no moral/ethical path to putting it back in.

This certainly highlights the problem. For context, I wasn't suggesting eugenics, although I suppose that's one way to enforce things (one I would never support, anyways).

Thought-experiment: Someone modifies their gametes for gene-drive extinction. [0]

Failure to disclose it to every potential partner and descendant would be a very serious kind of fraud or potentially child-abuse.

[0] https://www.scientificamerican.com/article/gene-drives-could...

> Isn't restricting people from reproducing based on their genetic material known as Eugenics?

We really need to stop being afraid of words. We don't have to use a neutral word to imply a whole range of ancillary things which we don't name.

There's nothing wrong, in principle, with eugenics. (Some bad guys many years ago were really into it for a specific use case, but so what?)

While humanity is in the learning phase of gene editing:

- High-trust societies can choose to say: "You can get whatever gene mods you want but you will be sterilized."

- In freedom societies you could simply ask your partner what modifications she has before you conceive kids with her. Screening and embryo selection would probably be available in such a world and individuals would make their choices.

> the freedom to perform "body hacking" on yourself. But, at least in this case, isn't it more like "generational hacking?"

The latter would certainly be much easier than the former. But especially people who don't work in bio dream about hand-wavy future biotech that does things like using a virus vector to spread a CRISP/CAS9 payload to all relevant cells in an adult body, and performs in vivo edits.

I'm extremely bearish on genetic tweaking as I think longterm unforseen consequences are essentially guaranteed, but I don't find this exact issue to be much of a problem. It's 100% legal to drink/smoke/etc to your heart's content while pregnant, so I think individual liberty still unambiguously wins here.

People like to compare CRISPR as scissors or cut/copy-and-replace, but people should think of it more as 'find and replace all'.

in that context, it's much more understandable why the tool has drawbacks.

Yeah, better analogy, but still a little flawed because that still makes it seem much better/precise than it is. It's "find and replace all" that might:

- occasionally match very similar, but not exact text

- will often subtly alter the text around the replacement area, either by deleting characters or altering others.

It’s more like pasting your genome into chatgpt and asking it to fix it

Sounds like my junior developer copy-pasting regular expression parts from stack overflow and pressing replace all while not really understanding what the funky character soup actually does. That fits your description pretty nicely

It's find-replace-all, but instead of being exact string matching, CRISPR uses somewhat imprecise RegEx.

Problems in repair pathways are underappreciated. The mutations causing breast cancer, and Huntington’s disease, for example. Advancements in treating those will be advancements in CRISPR.

Let's say you have a running program on a computer, and you figure out a way to swap out parts of its instructions / state in RAM while it is running. What are the odds of your swap causing problems? Now, what if the program is 100 - 1000x more complex than anything you have ever managed to create.

You may not remember GameSharks, but those things did you exactly what you suggest. As do most game cheat engines. Editing the state, directly in RAM, without the program's knowledge.

The next time something tries to use whatever memory or function it overroad, it would pick up your version instead.

I think this is more like introducing and RCE into your body, since CRISPR essentially modifies your executable code?

DNA is more like configuration for your molecural factories.

configuration is code.

speakig like a true "exe"

Not in this case. There is only a very restricted set of options and these can be interpreted in different ways by different machines that read it.

And merge conflicts are a bitch when you don’t have unit tests.

It's code, but not a program - there is no algorithm. Something resembling an algorithm is on the reader side.

by that logic, programs have no algorithms either, they have to be executed by the hardware.

There is a difference between a series of instructions that maps to machine code nearly directly, and a data format like an image, or factory templates.

The point about algorithms stands - it's not that programs don't contain any because a machine interprets them. I'm referring to the simple fact that program code can be expressed as a flow diagram - but try doing that with a CAD file. You can't, it's declarative in nature, not a series of steps and decisions.

You're trying to say that data isn't code? you haven't tried lisp, have you?

FYI doing this to a game is illegal in Japan, which kinda makes it seem like CRISPR should also be illegal in Japan.

I remember having a Game Genie, it worked sometimes. But like the article says. it would also make things glitchy and crash everything now and then.

Not surprising. They weren’t designed cheat codes. Folks would try stuff, sort of figure out what it did, and then publish it with a name.

Biology is already probabilistic, there are things going wrong with the body all the time. Most often this affects a single cell and is corrected by programmed cell death...

This is how many live-update and hot-patching systems work.

Some changes are easy and reliable; if you need to add a new condition or a few instructions, you can build your new basic block of instructions elsewhere and then atomically insert a jump to that basic block (which then jumps back to the correct post-insertion point).

Others are hard; if you need to add a new variable to some state that's being tracked, then you have to find a way to know when it's a safe time during execution to make that change, and you might have had to wrap accesses to that state in RCU gadgets.

It takes expertise, but it's doable and satisfying when you manage it!

Consider a scenario where you're editing a function:

  function foo() { return a*2.1^2+0.52/2 }

So you do a find-all regex "1.*5" and delete all matching occurrences (à la CRISPER) to get:

  function foo() { return a*2.2/2 }

But unbeknownst to you, the code is littered with a bunch of commented out versions of the same function you're trying to edit:

  /* function foo() { return a*1.5/2.1 } */
  /* function foo() { return a*1.95/2.4 } */

And now those commented out versions now become:

  /* function foo() { return a*/2.1 } */
  /* function foo() { return a*/2.4 } */

And now the whole program doesn't compile anymore--or your patients get Leukemia. Oops.

You can't parse DNA with regex¹.

¹ https://stackoverflow.com/questions/1732348/regex-match-open...

Excellent point. But it works for bacteria I guess, because viral dna is small.

And when it doesn't work... well, bacteria die all the time. Humans are less tolerant of random death.

So good, thanks for resurfacing this one!

There were some numbnuts that were calling themselves "Bio Hackers" and freely injecting themselves with random CRISPR injections. I wonder what's going to happen to them.

We should steer clear of modifying DNA directly until our understanding of biology is much much better.

For now, I think targeting RNA as an intermediate solution is the right approach.

I think there are plenty of diseases where people would be willing to roll the dice on low probability unintended consequences if it meant a massive 100% probability life improvement. This research is great because it reduces our uncertainty about the former option.

That said, I wouldn't wanna be the first person to do elective CRISPR therapy for something that I don't actually need (i.e. LASIK, etc).

Somatic cell modification is relatively safe. Or rather, not existentially dangerous to the species unless both extremely widespread and extremely deleterious.

How are we to learn how biology works better without modifying DNA directly? We learn through experimentation, not sitting upon a pedestal and pontificating for decades.

> Humans don't live for centuries! We can't wait for progress.

Of course editing genes with CRISPR can introduce other defects! Genetic sequences can serve multiple purposes, alternative splicing, polygenic traits etc. It's an exciting discovery with interesting repurpose applications but DNAs application is not just a linear read

I once got into a debate with a classmate in undergrad. He had seen a Ted Talk and was very worried that CRISPR was going to create a new race of designer babies. I tried to explain that even CRISPR wasn't there yet, but he was under the common misconception that CRISPR is just a text editor for the genome [1].

If we want to take the computer code analogy, CRISPR is not vim/emacs/nano/ect., it is sed -i 's///g' with greedy options on.

The 'g' option is what got the researchers here. I hypothesize that a future problem will be CRISPR targeting previous CRISPR edits since the targets are relatively conserved.

[1]: https://xkcd.com/1823/

ScholarlyArticle: "Gene editing of NCF1 loci is associated with homologous recombination and chromosomal rearrangements" (2024) https://www.nature.com/articles/s42003-024-06959-z

I wonder if there's a pattern to genes that are adjacent to the target ones, which is not typically found adjacent to the pseudo ones. In other words, can the "search term" be expanded to make it more selective and then take action on the target region within it.

We do this because reversing what caused the defect is way out of our league, just like how most advanced drugs work. The point is you take it anyway because you likely have 2 choice.

They have a targeted DNA edit that will fix the defect. They can't apply it cleanly.

Reversing the cause would require something nonsensical like editing their ancestor's DNA before they were born.

I see a whole bunch of comments here where people have misunderstood the issue.

The problem here is not that unanticipated outcomes arose from our intended fix.

The problem that the exact edit we planned isn't happening in the first place, because the search-replace tools aren't yet specific and reliable enough.

The genetic version of the Clbuttic Mistake?

https://thedailywtf.com/articles/The-Clbuttic-Mistake-

This is also not quite accurate. The edit is happening, but the problem is the same word appears three times in the same sentence, and the program can’t distinguish them.

Specifically, there are three copies of the gene, the active one and two inactive ones. Trying to only edit one, they ended up hitting ay least one of the others as well. That then caused misalignment and other issues for the rest of the neighborhood.

It's both. With 2 or 3 cuts, you have several pieces of DNA that recombine in whatever arrangement. So in addition to the disease not being treated, you have other issues.

When the sections are subsequently rejoined, entire gene segments may be misaligned or missing. The medical consequences are unpredictable and, in the worst case, contribute to the development of leukemia.

Does it matter which CRISPR or CRISPR-like method is applied; or is there in general a dynamic response to gene editing in DNA/RNA?

The article covers this and I think the title is a bit too general. It is a byproduct of how CRISPR works as it targets a specific sequence. In this case the sequence is also present in areas that were non-targeted. Essentially, the sequence was not unique so the process impacted other areas in unintended ways.

> Here we evaluated diverse corrective CRISPR-based editing approaches that target the ΔGT mutation in NCF1, with the goal of evaluating post-editing genomic outcomes. These approaches include Cas9 ribonucleoprotein (RNP) delivery with single-stranded oligodeoxynucleotide (ssODN) repair templates [11,12], dead Cas9 (dCas9) shielding of NCF113, multiplex single-strand nicks using Cas9 nickase (nCas9) [14,15], or staggered double-strand breaks (DSBs) using Cas12a16.

- "A NICER approach to genome editing with less mutations than CRISPR" (2023) [cas13d] https://news.ycombinator.com/item?id=37698196 :

"Prediction of on-target and off-target activity of CRISPR–Cas13d guide RNAs using deep learning" (2023) https://www.nature.com/articles/s41587-023-01830-8

In mice?

Changing code you don't understand can have effects you did not intend.

Who would have expected that?

They are more or less exactly describing the change they are making, indicating they do understand the code, they just can't edit it cleanly.

Oh, maybe I missed that we already completely understand how DNA works.

Yes, we do understand the mutations that we would target with something like crispr, where protein coding has been disrupted by a small error.

Ok, but since we don't understand everything about DNA, that can have unwanted side effects.

Are you making fun of me or do you really not understand that?

I'm not making fun of you, it's just that we do know what point mutations are and understand things well enough that there won't be unwanted side effects to fixing them when they are causing significant problems.

Not being able to cleanly fix just the mutation is a separate issue from that understanding.

No shit.

I would personally love to get a gene theraphy for an inherited auto-immune disease, but mankind just doesn't have 'the full picture' - yet. So i'll stick with known treatments.

I upvoted this because it's perhaps the most important research being conducted currently. But the seriousness of the issue is misleading.

CRISPR is revolutionary technology. Any side effects of its use require evolutionary improvements. The two are not the same.

With the decades of experience we have dealing with code and information, as well as the arrival of AI, I wish that people would stop obsessing about the difficulty of solving problems. The difficulty is gone. All problems are solvable now, it's just a matter of time, effort, and a tiny fraction of the money that used to be required.

But FUD seems to dominate these discussions. That's probably my biggest disappointment with, say, the rich and powerful who gatekeep funding and the media. It's even easier for them to solve these problems, or encourage healthy debate. But they seem to go out of their way to do the opposite. Like, if it weren't for wealth inequality, we could have been working on this stuff, and found the answers potentially years ago. I'll just never understand the so-called realists and how they hold us back in these times out of fear, which leads to false equivalence and generalization.

Reply on news.ycombinator.com