Very fast CRISPR on demand | Hacker News Reader

Very fast CRISPR on demand

science.sciencemag.org

66 pointsXzetaU831 comments

Will be interesting to see how ML and CRISPR converge. Two systems we don't fully understand creating explicit capability we desire and latent capability we don't expect.

I'm curious if there will eventually be some kind of biological incarnation of cellular automata that is useful for chemical synthesis or material fabrication. Clearly we have that with individual organisms now (e.g. insulin manufacture) but imagine seeding a matrix with bacteria that extrude carbon nanotubes under UV light or similar oddities.

Remember that CRISPR is just an editing technique, and we have the technical capabilities to synthesize genomes. To get to cellular automata, I think there needs to be much more focus on building rather than tweaking.

Though tweaking is very nice for eukaryotes, where we have no idea how to build and need to learn from small edits.

> we have the technical capabilities to synthesize genomes

Of very limited size (for now).

> Will be interesting to see how ML and CRISPR converge

I've been seeing similar comments a lot lately, and I don't understand why. As someone with a background in ML that is now studying biochemistry I'd be the first person to jump at something like that, but I'm just not seeing any big ML potential that directly has to do with CRISPR.

Is it just two hyped technologies being thrown in a pot? Is there something I'm missing?

As someone about to graduate with a PhD in biochemistry who has been programming for ~10 years and personally used CRISPR/Cas9 ... the parent comment gave me a good laugh.

That being said, I can see some applications to using ML to analyze the results of CRISPR/Cas9 screens or better predict functional editing sites - but that's not exactly exclusive to Cas9.

Exactly what I had in mind. ML has its place in bioinformatics (and is being applied to a huge amount of sub-problems there), but it almost always is an incremental improvement for that niche rather than transforming the whole landscape.

I'm talking precisely about what this guy worked on - https://news.ycombinator.com/item?id=23540564

ML provides predictive capabilities over large fugly datasets. CRISPR gives us the ability to test those predictions. These are both essentially commodity technologies at this point and not everyone is going to take a disciplined approach to experimentation.

I know you what you're missing. You gotta put AI in the Blockchain before adding CRISPER and applying for funding.

EDIT: And don't forget Kubernetes.

Serverless Gene editing!

CRISPRAAS

You laugh but...

https://www.genscript.com/crispr-services.html

https://canopybiosciences.com/custom-cell-line-engineering-2...

https://www.reprocell.com/gene-editing-services-i174

etc. etc.

This guy lifestreams some really interesting work - https://www.youtube.com/user/TheChemlife/videos

It's not complete though until someone rewrites the whole thing from the ground up entirely in rust and declares it now safe to actually use.

I've only heard of the two topics seriously discussed together in the context of directed evolution. Instead of using radiation to increase the rate of mutation, use (semi) randomized snippets injected via CRISPR with a feedback loop that includes your optimization function (i.e. increasing temperature range for a protein).

The conclusion was that the data would be far too sparse for the algorithm to really learn anything.

My Ph.D. work has used machine learning to model CRISPR editing outcomes, see https://www.maxwshen.com/

> Will be interesting to see how ML and CRISPR converge.

don't forget blockchain

Funny to see sales/marketing lingo in the title of a scientific article, but with an unexpected meaning.

I think they mean "on demand" is that the Cas9 proteins is activated upon activation of the light source. As the CRISPR/Cas9 complex is already in place due to the chemically boxed guide RNA, the way I understand it from briefly skimming this abstract is literally that you turn on the light when you want DNA cleavage, so I don't see anything weird about it, but maybe someone with more insight can correct me if I am wrong.

This is the terminology they introduced, very fast crispr on demand (vfCRISPR). not meant as a "competitive advantage" but as a description of its properties

(hence the title shouldn't have been edited)

Site is hugged to death already.

Archive snapshot can be found here: https://web.archive.org/web/20200616103722/https://science.s...

Can someone explain?

Here's the Perspective piece on the paper: https://science.sciencemag.org/content/368/6496/1180

still don't get it :D for dummies explanation?

I'm at the limit of my understanding myself, but here's what I think is being said:

CRISPR refers to the kind of DNA site that can be targeted for cleavage, while CAS9 is a protein that, paired with an RNA matching that site, can cleave wherever it encounters a matching sequence.

I think these guys are pairing a variation of the CRISPR approach with a laser. They pre-bond with the DNA, but the actual cleavage occurs when they shine a laser at the right frequency at the site. That lets them decide when and where on the DNA strand cleavage happens.

If anyone understands this better than me, I'd appreciate any corrections to my explanation ;P

I am fortunate to have had a colleague explain it to me but this is my understanding as well. Normally the cleavage would take time with possible undesired effects as a result of the process.

Close - from my quick skim here's my take

The RNA guide (sgRNA) is chemically caged here, so it has some molecular decorations that are removed with light. The sgRNA can't fully match the underlying DNA it's recognizing until those decorations have been removed with the laser, at which point Cas9 then cleaves the DNA strand.

Think of the sgRNA like a regex pattern. Cas9 is like the regex engine here (but instead of capturing a group, it cuts the sequence), and DNA is the search space. In this system, we essentially have a full match, and Cas9 is in position (having found the match) but it can't cleave yet, because the decorations are blocking the way. As soon as the decorations are removed by light, the sgRNA can move down into a full match, and Cas9 then cleaves.

Ah, thank you. So is the benefit of this system that the Cas9 sticks around for a while, and at any point one can cleave all the sites at once instead of cleaving each as the Cas9 bumps into it?

Even in standard CRISPR/Cas9 systems, Cas9 sticks around for quite a while - which is one of the issues in using it for actual medical treatments. You don't really want a programmable DNA cleavage engine roaming around in your nucleus...

The primary benefit here is that you can temporally synchronize Cas9. In standard systems, you introduce Cas9 into the cell, and you have no control over when it performs cleavage, since it's essentially a greedy cleavage mechanism - as soon as it finds a match, it cleaves.

This system basically introduces a semaphore, so that you introduce Cas9, allow it find a match and pause. You can then signal for cleavage to begin. As stated in the abstract, this is useful for e.g. studying the kinetics of DNA repair, since you can control when DNA damage (Cas9 cleavage) is happening.

Fascinating. Thank you so much for taking the time to explain further!

The Cas9 protein (from the CRISPR system) has 2 functions:

1) It binds to DNA based on where a template RNA strand matches DNA in a genome. That RNA strand has 2 pieces - a piece that attaches to the Cas9, and a piece that matches whatever sequence of DNA you want to bind.

2) It cuts the DNA it has bound to.

(The first property is very special to Cas9. The second property is actually pretty common. The common intention in using Cas9 is to cut DNA at a place that needs editing and hope the cell's own repair machinery fix it properly.)

Important to understand about biology/chemistry, things rarely happen "IFF and only IF", but more often, happen at a rate (i.e. 1 event per microsecond).

So better more true functions of Cas9 are actually:

1) Cas9 spends a larger fraction of time near its target sequence than it does around random sequences.

2) Cas9 cuts DNA ever 1/N timeperiods, and that might be sped up when it is attached to DNA.

So Cas9 should hang out around its target sequence, most of the time, and that the cutting event happens relatively randomly. So _in general_ Cas9 will cut at its site of interest, however, it will also cut elsewhere, purely statistically speaking. And if you want to be guaranteed a cut at your spot of interest, you're likely to get cuts elsewhere ("off-target" cuts to DNA. These are undesired as you really don't want random changes to your DNA that you're not asking for...).

----- now comes the paper -----

If you make it so Cas9 cannot cut because the homing device is kinda kinked or boxy or otherwise not quite right (but it is still able to home in on particular sequences of DNA), you can make it so Cas9 runs around and finds where its supposed to be, but just hangs out there; Function 1 still works, but function 2 is turned off. These scientists made a special version of RNA that only permits the Cas9 to work when light is shined. So the RNA absorbs the light from a laser, breaks a seal, and the Cas9 snips right then and only then (in a matter of seconds). And any Cas9 molecules that did not get light cannot cut - they just continue to hang out and don't do any damage that they otherwise might do, statistically.

If you're trying to cut a specific sequence and only a specific sequence, this increases the statistical chance you are correct by shortening the time period (and spatial localization) when Cas9 can actually take its second, cutting action.

CRISPR systems are molecular "scissors" you can use to cut a particular piece of DNA right where you want to, which is useful when doing gene editing. Usually, you put the scissors inside the cell, and it just sort of goes and cuts on its own timeline, you can't really control it. Here, they've designed a pair of scissors that will go right to where you want to cut, but then just hang out there not doing anything until you give the signal. When you give the signal (activation by light), the scissors make the cut "on demand."

So before, if you want to study/control the processes that go on after that cut, you'd be a bit out of luck since you didn't know when the cut would happen to start experimenting, it would be a hard messy problem. Now, you can just trigger the cut to happen right when you're ready for it.

Thanks

Reply on news.ycombinator.com