The obvious favourites: - Drug development - Plastic recycling But I want to see cheap enzymes for common industrial processes, so we can upcycle plant waste or change the properties of construction timber without nasty chemicals.
@@vlazurenko What I'm advocating is a non-authoritarian and fully transparent approach where decisions are made together, based on scientific expertise, so as to maximize access within Earth's carrying capacity, ensuring that everyone's needs are met. Resources are allocated based on desirability and availability in the most fair and sensible way possible via dedicated open-source algorithms. Again, no police is required, only comprehensive rehabilitation.
@@vlazurenko Crime, dominance and abhorrent are nothing but symptoms of unmet needs. and existing criminals can actually be resensitized. We need no prisons, judges and laws. Hunter-gatherers had none of those after all, not even money and they thrive for 300k years peacefully by sharing everything, until agriculture created dense populations where scarcity prevailed and forced people to compete. That said, we're naturally wired to collaborate, not compete.
Me too. Not even close to being in that field. But. We all are made of biology. Advances in it are amazing, and this may be huge? Can't say i understand but it sounds like it.
@@ristopoho824 I only have the most basic layman's sense of what this all may mean for medicine, but it seems like it and everything else being developed and implemented that is similar to it will have a massive impact on medicine before very long. Five years ago, even top experts in the field weren't expecting anything like what has happened since (not just DeepMind's projects - many programs) for decades.
It's not but we are getting closer. Ninety percent sounds good but that's a lot of error in a complex non-linear system. Morever, we need to a model of dynamical protein behaviour. Proteins are not static elements, they are a dynamic structure, continuously forming different configurations. What this model is attempting to solve is the most stable configuration or lower-energy configuration. Proteins exist in a distribution of configurations that can be modulated intermolecular interactions and the bonding of other molecules to the protein.
I don't know what "For seven of the targets, between 9% and 88% of the designs tested in the wet lab were experimentally verified as successful binders" means and I can't find the source. 9 - 88% seems quite a range, and this begs the question what happened with the designs for the other targets.
they had seven targets, and made multiple binders for all seven. One of the targets had a 88% success rate (which was the highest of the seven). The lowest success rate was 9%, and the last 5 targets had a success in the range of 10%-40%. But yeah, 9% to 88% is quite a range, but that's very usual for synthetic biology research.
Go to the paper (in the video description) and look at Table 1 (use Ctrl+F to find it if needed). The columns are the eight target proteins that success was tested for. In each row, you have the success rate for some source of designs (AlphaProteo first, followed by RFDiffusion, followed by other computational design methods) followed in brackets by the number of designs tested from that source. Of the eight target proteins, in the eighth column, no successful binders were found for the target protein, although the authors didn't find any successful designs to compare to. In other columns, the success rate for the remaining seven proteins varied from 9-88% for different target proteins. Although 9% is not much, the table shows the success rate for designs from other sources (excluding RFDiffusion) for that target protein was only up to 0.07% from 14,982 tested designs. 'Success' is defined as a design that exhibited 'measured binding' with the target protein, so it doesn't always mean the design was particularly effective; the second part of the table shows the 'binding affinity', showing how effective the designs were when they were successful (compared to the most effective existing computational design in the bottom row). As in the video, for the second part of the table, a lower score is better.
Prion diseases are among the scariest things out there. I hope work like this goes some way to finding treatments if not making conditions like CJD, Parkinson's or FFI a thing of the past.
Prion disease is misfolded proteins, these people are playing with fire and profit will make them bring to market something that will make them feel like gods and we will learn the harsh way that they are not.
That's exactly what I want: a protein that binds to misfolded molecules that can repair their shape or something like that. That would be a cure for my mother's Parkinson (or at least, it would stop its progression).
Incredible potential here to design new classes of proteo-antibiotics, -proteins to inhibit or kill specific bacteria, all without targeting the good bacteria or off target side-effects. Also, the same technology could be used for rapid diagnostics of microbes, - both viral and bacterial, possibly even down to the strain level. Congrats to the team at DeepMind! It is hard to overstate the impact of this monumental effort.
You mean openai should put off public releases until they've done the proper research and made sure their products work and are safe and beneficial? Unpopular opinion there mate.
@@bfyrth This is generative AI. It generates proteins that can bind to other proteins. Deep learning and generative AI are not mutually exclusive, indeed I believe all generative AI uses deep learning.
Keep in mind that everything must always be experimentally verified and tested. However, the more tools we have, the better. Perhaps one day designing a custom protein will be as simple as designing a DNA primer for PCR. We really need protein dynamics with covalent modifcations as well. Much of the interesting stuff happens via regulatory mechanisms like activing some protein binding site or blocking it, for example. It would be nice to predict the effects of protein modifications, especially if we get to the point where we can run these affinity calculations against an entire proteom at once. Biology is moving towards Star Trek, one step at a time. I can't wait to design entire regulatory systems and have a company just ship the required components to you!
That's why this is so important - *everything* must be experimentally verified and tested. So if we can generated much better guesses to test, that testing will be far far more fruitful. Testing is so expensive and acts as an effective block to research in most areas because our guesses are bad yet take weeks to months to years to discover that they're useless. Getting that success rate up is of paramount importance.
I generally treat DeepMind stuff with a grain of salt. If you remember that AI that found "100,000 new compounds for scientists!" scientists tried them out and it turns out the vast majority of them were utterly useless. Google also has a history of faking numbers or demoes (the gemini demo). Given this, and the basically non-answer of "well the success rate is between 9 and 80%" (???????) I put very very very little hope in this being some huge breakthrough. A great starting point definitely, but it doesn't completely revolutionize the field.
@@yobabadakong8137 Yeah this is the typical "balanced news report" or RUclips contrarian schtick but it's not even close to accurate. We already have very advanced bioweapons.
I would first use it to perfect enzymes that could replace opioid pain killers and adhd stimulants with more effective versions, much longer duration, but without those massive side-effects Next I would try to find out can they be used to neutralize auto-immune disease anti-body attacks in a more contained and effective manner. Then I'd try to find out why some organ donors reject some organs so aggressively, while others it tolerates. And see if we could a) match people with organs they won't reject b) possibly learn to inhibit only that rejection, without requiring broad range immuno-suppressors that can be extremely dangerous by themselves
And once you patent them for the good of humanity, who will spend the several billion dollars developing that open-source protein into a usable drug, getting it through expensive clinical trials, then marketing it and withstanding class-action suits from ambulance chasers, all so someone else can come along and copy their homework and there's not a damn thing they can do about it?
Seeing how it all works in practice, it will be a battle of big pharma with patent trolls. Patent trolls will patent random molecules and then not even produce them, waiting for somebody to use that molecule so they can sue them and make their money. For big pharma it will be easier to gain monopoly, because now when they work on a target, they will discover all the other possible drugs for that target and patent them, and then inflate the price many orders of magnitude for one product that they make. I'm a huge pessimist when it comes to the current system. Really, currently, the only way to have affordable drugs is to order synth of them yourself from China through a gray market. And sometimes that's the only way, due to patents and regulations, and those molecules simply not being available.
@@utkuaYes, they will use other people's work to discover new drugs, then patent them, and then use money gained from the monopoly to lobby even more regulation to make those drugs even more unaccessible so they can maintain their monopoly. That is currently how it works, and I don't see anything that can change that.
I have been suffering from multiple sclerosis for 22 years, since I was 21 years old, which is more than half of my life. I always hope for solutions to this disease, but now I am in a wheelchair. It feels like I am a prisoner in my own body. I want to enjoy my life without suffering for so long. Can you please find a cure for these diseases?
Trying to edit the post but it doesn't let me. I meant to make it more clear that the search engine. bioenergetic life. is what it goes by. That is the literal name for this search engine
Alpha fold family is one of my favorite papers. It saves so many time and effort with medical creation. It is literally became silent revolution, that can be used widely, not just shine beautiful.
I am currently in a lab as an undergraduate that works on the applications of these AI designed binders and seeing this makes me super happy as even with the previous processes we are able to get some fantastic results. I am going to have to read the paper for some more of the details as I would like to see if there are any downsides to this new process, but it still looks like a massive leap forward.
Yeah. F*ck that ending up sh*tting myself in diapers and being spoonfed. The projections for neurodegenerative disorders and the growth in nursing homes in most of the world and the exhaustion of pension plans by only 2050 is mind-boggling and so many governments are just pretending it doesn't exist. Well, look at Japan now as an example. For some reason _their_ government is investing in anti-aging research. And it does suck. I blew out my back when I was young and dumb at a brutal job ONE summer so now that I'm middle-aged I'll deal with it for...for what? Forever?!? The next 40 or 50 years? F*ck that sh*t. We do need to fix it.
This is a problem so complex that I've never thought I would have seen such a rapid development in my lifetime! I hope it could change completely how we approach the development of new drugs and treatments. And honestly, it's moments like this that remind me to be grateful of the time I live in. 🙏
Thatll mean we will miss out on all of the money for cancer, thats a huge amount, if a cure would exist we would miss out on all that money, its better for society for there not to be a cure
This is a significant development that should be widely reported in the mainstream media, but it is not. This means that humanity is one step closer to eliminating diseases for good. However, pharmaceutical companies will find ways to delay progress and then profit from it.
That's because mainstream media is just corporate PR. You shouldn't expect to hear accurate reporting on scientific matters unless it is tied to a consumer product or favored political party's talking points. The only reason the news reports on medical advances is because they are paid to do so by the companies that sell those technologies (e.g. the Bill and Melinda Gates Foundation pays news companies like CNN and PBS to report on pharmaceuticals that the foundation develops through its own subsidiaries).
The only reason mainstream news programs report on medical advances is because their advertisers and investors are marketing that technology through those news programs. They are paid advertisements disguised just enough to evade scrutiny from the FTC.
I want to thank you for your positive videos, with an optimistic approach all throughout! I am excited to experience the future of technology alongside you Dr. Karoly Zsolnai-Feher. ( had to search up how to spell the name :D )
...but why are we promoting tools like this that are not open-source and not reproducible science? The same thing goes for AlphaFold3. (I'm all for companies being able to have commercial products, but let's stop giving them papers in top science journals and acting like this is anything but an advertisement.)
I'm a biologist specialized in proteins and AI. And I can say that, while ALPHAPROTEO is awesome,IS NOT A GIFT FOR HUMANITY, instead, is only a paid commercial by deepmind as well as AlphaFold3. Deepmind doesn't release the code for both tools, and as a scientist, we need that in order to produce science. A webserver like the implemented in AlphaFold3 is not useful. And even more, it limits the capabilities that the model could reach. Other observation is that DeepMind doesn't solve the protein folding problem. That's a common error. They solve the structure prediction problem posed by the CASP challenge. Protein folding implies known the dynamic behavior of proteins along nano to microseconds. For that we use Quantum mechanics to model the behavior of the atoms una protein through the time
Yes. This is an ad disguised as a neutral video and the guy's vocal intonations are like sandpaper. However, the code for AlphaFold 3 and AlphaProteo will effectively be open source anyway about a year from now. DeepMind can only keep it locked down before a bunch of people working furiously right now succeed in cracking it and slightly changing it to avoid legal issues and putting it online truly for free. Which DeepMind knows. It is trying to get out ahead with partnerships with certain drug companies is all it can do.
Is this an actual scientific advance? No ''this isnt quite proven yet but it shows promise''. Just an actual solved thing?!??! Holy shit i thought we stoppe doing those, YESSSSSSSSS
Mind-blowing video! 🧠💡 AI's greatest promise? Cracking the code on major diseases and unlocking human biology. So grateful for this potential health revolution. Here's to a future where AI helps us all live longer, healthier lives! 🌟 #AIforHealth
What I would use it for: 1. Make an enzyme that dissolves artery plaque 2. Make a protein that kills any cell around it, but that only activates in the presence of other proteins emitted by cancerous cells.
Is this not a future possibility for human DNA evolution? Imagine getting the ones that tardigrades have (radiation immunity + more radiation immunity)
I've been working in laboratory sciences and studying ML for years, and I have been waiting for this moment. This tool, and later iterations of it, will allow us to produce designer proteins that adhere to receptors on viruses and prions blocking them from connecting with human cells. As we use this tool and then confirm the predictions through experimentation, we will build larger datasets to train the next iteration on.
It's always important to remember - this is the worst this program and all of those like it will ever, or can ever, be. I keep forgetting this. AlphaFold 2 was impressive, then AlphaFold 3 all of 3.5 years later was a huge improvement. This is much more ambitious even than AlphaFold, and it was released four months after 3. Biology is a ridiculously hard field, so it's nice to see AI starting to crack it. Biology is to this century what physics was to the 20th. Except far more impactful for our lives. I always thought it was odd than we could see the background radiation of the Big Bang or erase cities in an instant but couldn't fix basic medical issues. Something seemed way off. But really, we needed to build the massive foundation in physics first. And maaaybe our insane priorities involving blowing vast sums on wars and Cold Wars and banning all psychedelic research everywhere in the world half a century ago and allowing GPs and drug companies to get away with literal murder regarding opiates and benzos and so forth might have also had something to do with it...
I wonder if a similar technology could be used to run prebiotic chemistry simulations? Not in an attempt to replicate certain conditions (like volcanic vents), but just to find out if it could find feasible paths to construct things out of simple organic molecules (like amino acids, nucleotides, and lipids).
This is the stuff that makes Science Fiction a reality. Combine this with Orbital Materials and i get dizzy and need to sit down. I personally can't wait for the mapping the new material 'spaces' and learn insights in reliability, toxicity, environmentally friendly and so on. On the bio side a streamlined series of AI's to detect, design and prepare samples and jump straight into another series of AI's to optimize testing. I bet we can shave off weeks for any need to get something to entire population levels. Exciting times indeed
@@cslearn3044 you're right, your 90 year old grandma in the dementia ward will be getting it first decades before you know it she will be evolved far past anything you could ever imagine...
i remember i heard that the protein structures can be seen by very complicated methods but it cost soo much money just to get one protein structure and then see if it can help to certain cases that be able to have a very good approximation using computers is so fucking awesome as it can lead to analyze structures and then see if can be used in certain scenarios without trial and error that cost millions and millions of dollars without even know if will help.
DeepMind's AlphaProteo AI is hailed as a major breakthrough for humanity, offering revolutionary insights into protein structures and biological processes. This advancement promises significant improvements in drug discovery and understanding complex diseases.
At least a part of that future will (should) reach most of us for free. It must be a right... To health, To education, To improve, With AI. It must not be only for the rich in developed countries. *We can't let that happen* .
even tho i dont understand how do they go from the computer version to actually creating a useful protein in the lab this sounds as groundbreaking as when DNA human sequence was finally "decoded"
To all those hesitant about what 9 to 88% success rate means, it changes your screening effort from thousands or hundreds of thousands of candidates to like 100 and years to a bunch of months, even one month could be enough if you ask. It reduces the discovery cost to like 0.1 to 1% of the original cost. Clinical trials should remain the same, it takes years and are expensive, but manufacturing of proteins instead of fancy small molecules is more scalable and easier to distribuye in CDMO's. As well, protein risk assesment is relatively easier than that of small molecules.
Interesting, but 7 out of 8 tests being between 9-88% successful sounds very much like random results to me. Maybe fairly accurate in specific cases but overall not as impressive as you make it sound.
Do not forget that this can be used in the opposite direction. The moral and moral level of the ruling "elites" makes us think first of all about this and about contacts with military departments.
Been hearing of Deepmind science-domain breakthroughs for five years now and no practical application follow ups. They really need to market practical successes as well to showcase the amazing work
Google don't market anything quickly. Google had LLMs around the same time OpenAI did (see Palm models), but it took OpenAI marketing the technology as ChatGPT for them to wake up and then release Bard/Gemini.
In real life proteins have more than one 3D structure. Varies with pH or even electric fields of nearby molecules. I think this is why the practical wet lab experiments generated such a range. It is not the inverse of the protein folding problem, but the inverse of the protein folding problems. And AlphaFold before only generated one 3D structure. Also many f proteins combine with metals (zinc fingers, iron in hemoglobin) to be functional. How does this work with non-amino acid componants and resulting shake and charge distribution changes?
And people think Google is behind on AI for some reason. Waymo and DeepMind are the two most successful real world AI projects in existence, both Alphabet companies.
great! finally we can simulate using propability all the possible proteins and what they do! who needs alien worlds proteins now eh? well maybe those never had proteins 😅
There may be a lot of serious diseases that won’t be solved by a single protein or drug, there could be some that will. And it will increase our knowledge base that will help bring some of the more difficult diseases within range of a cure further down the road.
1:40 Including, like, cures for diseases and pretty much all toxins produced by animals/plants (though many of these are more complex than a single protein, it's a big step)
The wet lab verification was comparable to current SOTA methods - does this mean it’s not a medical breakthrough yet or does this approach bring major speed or cost improvements over those other best methods?
i think this approach could bring major speed and cost improvement, and reducing the time it needs to verify it, thus this is technically an advancement, they're hyping it because they don't want to look bad to the investor/shareholder. this approach is more novel i suppose, and it has not been well tested, so we can't currently know if it's reliable or not, i have a fair amount of skepticism here, but the potential impact of this technology is undeniably immense in the long term.
The obvious favourites:
- Drug development
- Plastic recycling
But I want to see cheap enzymes for common industrial processes, so we can upcycle plant waste or change the properties of construction timber without nasty chemicals.
And bioweaponry of course. As long as we keep living by trade and competition instead of sharing and collaboration, we're doomed
In the meantime, pharmaceutical companies will find ways to delay progress and then profit from it
@@ziad_jkhan average collectivization / famine / secret police enjoyer detected
@@vlazurenko What I'm advocating is a non-authoritarian and fully transparent approach where decisions are made together, based on scientific expertise, so as to maximize access within Earth's carrying capacity, ensuring that everyone's needs are met. Resources are allocated based on desirability and availability in the most fair and sensible way possible via dedicated open-source algorithms. Again, no police is required, only comprehensive rehabilitation.
@@vlazurenko Crime, dominance and abhorrent are nothing but symptoms of unmet needs. and existing criminals can actually be resensitized. We need no prisons, judges and laws. Hunter-gatherers had none of those after all, not even money and they thrive for 300k years peacefully by sharing everything, until agriculture created dense populations where scarcity prevailed and forced people to compete. That said, we're naturally wired to collaborate, not compete.
I am not in biology, and I think I don't need to be in biology to appreciate this kind of work.
Me too. Not even close to being in that field. But. We all are made of biology. Advances in it are amazing, and this may be huge? Can't say i understand but it sounds like it.
@@ristopoho824 Sounds like they previously targeted a rat with a bomb, now they can target with a sniper rifle. Or navigating by stars vs GPS.
@@ristopoho824 I only have the most basic layman's sense of what this all may mean for medicine, but it seems like it and everything else being developed and implemented that is similar to it will have a massive impact on medicine before very long. Five years ago, even top experts in the field weren't expecting anything like what has happened since (not just DeepMind's projects - many programs) for decades.
> protein folding is now a solved problem
Never thought I'd hear those words in my lifetime.
90% solved, lol.
Yeah he said “mostly”
Kinda scary really, it could be used to weaponize prions in the wrong hands
It's not but we are getting closer. Ninety percent sounds good but that's a lot of error in a complex non-linear system. Morever, we need to a model of dynamical protein behaviour. Proteins are not static elements, they are a dynamic structure, continuously forming different configurations. What this model is attempting to solve is the most stable configuration or lower-energy configuration. Proteins exist in a distribution of configurations that can be modulated intermolecular interactions and the bonding of other molecules to the protein.
you mean protein misfolding* ?
I don't know what "For seven of the targets, between 9% and 88% of the designs tested in the wet lab were experimentally verified as successful binders" means and I can't find the source. 9 - 88% seems quite a range, and this begs the question what happened with the designs for the other targets.
2:36
I take this to mean that the ratio of successful designs varies depending on the target, which seems reasonable, some targets are harder than others.
they had seven targets, and made multiple binders for all seven. One of the targets had a 88% success rate (which was the highest of the seven). The lowest success rate was 9%, and the last 5 targets had a success in the range of 10%-40%. But yeah, 9% to 88% is quite a range, but that's very usual for synthetic biology research.
@@arunkumar_ra that's theoretically, those numbers don't necessarily reflect the experimental ones
Go to the paper (in the video description) and look at Table 1 (use Ctrl+F to find it if needed). The columns are the eight target proteins that success was tested for. In each row, you have the success rate for some source of designs (AlphaProteo first, followed by RFDiffusion, followed by other computational design methods) followed in brackets by the number of designs tested from that source. Of the eight target proteins, in the eighth column, no successful binders were found for the target protein, although the authors didn't find any successful designs to compare to. In other columns, the success rate for the remaining seven proteins varied from 9-88% for different target proteins. Although 9% is not much, the table shows the success rate for designs from other sources (excluding RFDiffusion) for that target protein was only up to 0.07% from 14,982 tested designs. 'Success' is defined as a design that exhibited 'measured binding' with the target protein, so it doesn't always mean the design was particularly effective; the second part of the table shows the 'binding affinity', showing how effective the designs were when they were successful (compared to the most effective existing computational design in the bottom row). As in the video, for the second part of the table, a lower score is better.
Prion diseases are among the scariest things out there. I hope work like this goes some way to finding treatments if not making conditions like CJD, Parkinson's or FFI a thing of the past.
Prion disease is misfolded proteins, these people are playing with fire and profit will make them bring to market something that will make them feel like gods and we will learn the harsh way that they are not.
That's exactly what I want: a protein that binds to misfolded molecules that can repair their shape or something like that. That would be a cure for my mother's Parkinson (or at least, it would stop its progression).
This is basically the same problem that occurs in Alzheimer's and Parkinson's diseases too.
I assume FFI doesn't stand for foreign function interface. What does it stand for?
@@lionelt.9124 Fatal Familial Insomnia.
enterprise-ai AI fixes this. DeepMind AlphaProteo: A Gift!
Incredible potential here to design new classes of proteo-antibiotics,
-proteins to inhibit or kill specific bacteria, all without targeting the good bacteria or off target side-effects.
Also, the same technology could be used for rapid diagnostics of microbes, - both viral and bacterial,
possibly even down to the strain level.
Congrats to the team at DeepMind! It is hard to overstate the impact of this monumental effort.
DeepMind is what OpenAi should be
You mean openai should put off public releases until they've done the proper research and made sure their products work and are safe and beneficial?
Unpopular opinion there mate.
People already start to believe that LLMs have intelligence and need to be treated well lol.
Very appreciated, Openai!
@@alansmithee419 Maybe an unpopular opinion, but most likely a commercially infeasible one.
two very different AI technologies, 'deep learning' vs 'generative ai'
@@bfyrth
This is generative AI. It generates proteins that can bind to other proteins.
Deep learning and generative AI are not mutually exclusive, indeed I believe all generative AI uses deep learning.
Keep in mind that everything must always be experimentally verified and tested. However, the more tools we have, the better. Perhaps one day designing a custom protein will be as simple as designing a DNA primer for PCR. We really need protein dynamics with covalent modifcations as well. Much of the interesting stuff happens via regulatory mechanisms like activing some protein binding site or blocking it, for example.
It would be nice to predict the effects of protein modifications, especially if we get to the point where we can run these affinity calculations against an entire proteom at once. Biology is moving towards Star Trek, one step at a time. I can't wait to design entire regulatory systems and have a company just ship the required components to you!
That's why this is so important - *everything* must be experimentally verified and tested. So if we can generated much better guesses to test, that testing will be far far more fruitful. Testing is so expensive and acts as an effective block to research in most areas because our guesses are bad yet take weeks to months to years to discover that they're useless. Getting that success rate up is of paramount importance.
I generally treat DeepMind stuff with a grain of salt. If you remember that AI that found "100,000 new compounds for scientists!" scientists tried them out and it turns out the vast majority of them were utterly useless. Google also has a history of faking numbers or demoes (the gemini demo). Given this, and the basically non-answer of "well the success rate is between 9 and 80%" (???????) I put very very very little hope in this being some huge breakthrough. A great starting point definitely, but it doesn't completely revolutionize the field.
This is incredible! This will save countless lives
Just gotta keep a watch out for those chemical weapons and bioengineered biological plauges, but yeah, this is a huge leap forward.
For the record, bioweapons are already extremely advanced, terrifying and banned (for good reason).
@@yobabadakong8137 Yeah this is the typical "balanced news report" or RUclips contrarian schtick but it's not even close to accurate. We already have very advanced bioweapons.
thanks to deepmind for changing the way biotech is done
I would first use it to perfect enzymes that could replace opioid pain killers and adhd stimulants with more effective versions, much longer duration, but without those massive side-effects
Next I would try to find out can they be used to neutralize auto-immune disease anti-body attacks in a more contained and effective manner. Then I'd try to find out why some organ donors reject some organs so aggressively, while others it tolerates. And see if we could a) match people with organs they won't reject b) possibly learn to inhibit only that rejection, without requiring broad range immuno-suppressors that can be extremely dangerous by themselves
So now the race is on to spam the AI, find as many binders as possible and then patent them for the good of humanity, instead of for private Corp.
Most of the world does not care about patents, patents only slow down the progress in the country they are filed in these days.
And once you patent them for the good of humanity, who will spend the several billion dollars developing that open-source protein into a usable drug, getting it through expensive clinical trials, then marketing it and withstanding class-action suits from ambulance chasers, all so someone else can come along and copy their homework and there's not a damn thing they can do about it?
@@RickinICT They are using other people's work and public funding to make those research. You can bet they are using open source software as well.
Seeing how it all works in practice, it will be a battle of big pharma with patent trolls. Patent trolls will patent random molecules and then not even produce them, waiting for somebody to use that molecule so they can sue them and make their money. For big pharma it will be easier to gain monopoly, because now when they work on a target, they will discover all the other possible drugs for that target and patent them, and then inflate the price many orders of magnitude for one product that they make.
I'm a huge pessimist when it comes to the current system. Really, currently, the only way to have affordable drugs is to order synth of them yourself from China through a gray market. And sometimes that's the only way, due to patents and regulations, and those molecules simply not being available.
@@utkuaYes, they will use other people's work to discover new drugs, then patent them, and then use money gained from the monopoly to lobby even more regulation to make those drugs even more unaccessible so they can maintain their monopoly. That is currently how it works, and I don't see anything that can change that.
I have been suffering from multiple sclerosis for 22 years, since I was 21 years old, which is more than half of my life. I always hope for solutions to this disease, but now I am in a wheelchair. It feels like I am a prisoner in my own body. I want to enjoy my life without suffering for so long. Can you please find a cure for these diseases?
May I ask what kind of treatments have you tried so far?
@@mmcmmc999 almost all available
Please pleaae please consider a carnivore diet, i guess you dont have alot to lose. Look up dr ken berry and anthony chaffee. Please try it
It feels like when I search on bioenergetic life there are some useful bits of info. It is a Raymond Peat search engine for many topics
Trying to edit the post but it doesn't let me. I meant to make it more clear that the search engine. bioenergetic life. is what it goes by. That is the literal name for this search engine
I guess I'm smart enough to know I'm stupid
That's the first step towards progress.
Weird place to be stuck in but relatable.
I would use it to heal Ankylosing Spondylitis … to make my wife feel better 5:12
Basically any really hard disease. Can't come soon enough.
Thank you to Google, Google Deepmind and everyone involved in this for sharing this freely to the world!
Alpha fold family is one of my favorite papers. It saves so many time and effort with medical creation.
It is literally became silent revolution, that can be used widely, not just shine beautiful.
I am currently in a lab as an undergraduate that works on the applications of these AI designed binders and seeing this makes me super happy as even with the previous processes we are able to get some fantastic results.
I am going to have to read the paper for some more of the details as I would like to see if there are any downsides to this new process, but it still looks like a massive leap forward.
Long COVID, nowadays. I'm literally emotionally crying for that. Thanx for sharing !!
What a time to be alive!
Well, the dream of proteomics is right on the cusp of being realized. It's definitely an exciting time to be alive.
I can't even imagine what we are going to discover two more papers down the line!
Come on hurry up AI! Aging is horrible We need to fix it already.
Yeah. F*ck that ending up sh*tting myself in diapers and being spoonfed. The projections for neurodegenerative disorders and the growth in nursing homes in most of the world and the exhaustion of pension plans by only 2050 is mind-boggling and so many governments are just pretending it doesn't exist. Well, look at Japan now as an example. For some reason _their_ government is investing in anti-aging research.
And it does suck. I blew out my back when I was young and dumb at a brutal job ONE summer so now that I'm middle-aged I'll deal with it for...for what? Forever?!? The next 40 or 50 years? F*ck that sh*t. We do need to fix it.
This is a problem so complex that I've never thought I would have seen such a rapid development in my lifetime!
I hope it could change completely how we approach the development of new drugs and treatments.
And honestly, it's moments like this that remind me to be grateful of the time I live in. 🙏
What an amazing time to be alive!
This is actually very good news. I'm impressed.
If this leads to a cure for cancer I’m all for it
As long as you are ok with it 😅
Cancer is 200 problems in a trenchcoat.
Ever seen the “hitler cures cancer” video on RUclips?
There isn't gonna be cure mate, you know how much money they would loose if there was a cure??
Thatll mean we will miss out on all of the money for cancer, thats a huge amount, if a cure would exist we would miss out on all that money, its better for society for there not to be a cure
Incredible advancement! AlphaProteo is setting a new bar for AI-driven innovation in medicine.
This is a significant development that should be widely reported in the mainstream media, but it is not. This means that humanity is one step closer to eliminating diseases for good.
However, pharmaceutical companies will find ways to delay progress and then profit from it.
Why cure diseases when you can profit from them?
It's not reported because it's not commercialized. The GPT technology existed for years and wasn't reported on until OpenAI marketed it as ChatGPT.
That's because mainstream media is just corporate PR. You shouldn't expect to hear accurate reporting on scientific matters unless it is tied to a consumer product or favored political party's talking points. The only reason the news reports on medical advances is because they are paid to do so by the companies that sell those technologies (e.g. the Bill and Melinda Gates Foundation pays news companies like CNN and PBS to report on pharmaceuticals that the foundation develops through its own subsidiaries).
Why? They want to make a ton of money too. And it will bring R&D costs way down.
Not _everything_ has to have a contrarian take, you know.
The only reason mainstream news programs report on medical advances is because their advertisers and investors are marketing that technology through those news programs. They are paid advertisements disguised just enough to evade scrutiny from the FTC.
I want to thank you for your positive videos, with an optimistic approach all throughout! I am excited to experience the future of technology alongside you Dr. Karoly Zsolnai-Feher. ( had to search up how to spell the name :D )
That is perfect spelling right there. Great work and thank you so much for being so kind! 🙂
...but why are we promoting tools like this that are not open-source and not reproducible science? The same thing goes for AlphaFold3. (I'm all for companies being able to have commercial products, but let's stop giving them papers in top science journals and acting like this is anything but an advertisement.)
0:58 protein folding is still not a "solved" problem
No. Just much better at prediction, which is still astounding. 'Solved' is inaccurate, however, and other videos avoid this. Because this is an ad.
This thing will be able to generate a ton of bioweapons at the press of a button. Neat.
I'm a biologist specialized in proteins and AI. And I can say that, while ALPHAPROTEO is awesome,IS NOT A GIFT FOR HUMANITY, instead, is only a paid commercial by deepmind as well as AlphaFold3.
Deepmind doesn't release the code for both tools, and as a scientist, we need that in order to produce science. A webserver like the implemented in AlphaFold3 is not useful. And even more, it limits the capabilities that the model could reach.
Other observation is that DeepMind doesn't solve the protein folding problem. That's a common error. They solve the structure prediction problem posed by the CASP challenge. Protein folding implies known the dynamic behavior of proteins along nano to microseconds. For that we use Quantum mechanics to model the behavior of the atoms una protein through the time
Yes. This is an ad disguised as a neutral video and the guy's vocal intonations are like sandpaper. However, the code for AlphaFold 3 and AlphaProteo will effectively be open source anyway about a year from now. DeepMind can only keep it locked down before a bunch of people working furiously right now succeed in cracking it and slightly changing it to avoid legal issues and putting it online truly for free. Which DeepMind knows. It is trying to get out ahead with partnerships with certain drug companies is all it can do.
As a bioinformatician, I can confirm that this is absolutely groundbreaking
This research has the power to save as much people as indoor plumbing and germ theory.
Truly, what a time to be alive!!
Deepmind's protein models are my favourite AI models
Is this an actual scientific advance? No ''this isnt quite proven yet but it shows promise''. Just an actual solved thing?!??!
Holy shit i thought we stoppe doing those, YESSSSSSSSS
Mind-blowing video! 🧠💡 AI's greatest promise? Cracking the code on major diseases and unlocking human biology. So grateful for this potential health revolution. Here's to a future where AI helps us all live longer, healthier lives! 🌟 #AIforHealth
Thanks for making such amazing research understandable to dummies like us!🧬
What I would use it for: 1. Make an enzyme that dissolves artery plaque 2. Make a protein that kills any cell around it, but that only activates in the presence of other proteins emitted by cancerous cells.
Dr. Károly Zsolnai-Fehér always shares the good news on research and does so with an optimistic viewpoint, thanks for this
Is this not a future possibility for human DNA evolution? Imagine getting the ones that tardigrades have (radiation immunity + more radiation immunity)
I'm a simple man, I see a two minute papers upload, I click
I've been working in laboratory sciences and studying ML for years, and I have been waiting for this moment. This tool, and later iterations of it, will allow us to produce designer proteins that adhere to receptors on viruses and prions blocking them from connecting with human cells. As we use this tool and then confirm the predictions through experimentation, we will build larger datasets to train the next iteration on.
It's always important to remember - this is the worst this program and all of those like it will ever, or can ever, be. I keep forgetting this. AlphaFold 2 was impressive, then AlphaFold 3 all of 3.5 years later was a huge improvement. This is much more ambitious even than AlphaFold, and it was released four months after 3.
Biology is a ridiculously hard field, so it's nice to see AI starting to crack it. Biology is to this century what physics was to the 20th. Except far more impactful for our lives. I always thought it was odd than we could see the background radiation of the Big Bang or erase cities in an instant but couldn't fix basic medical issues. Something seemed way off. But really, we needed to build the massive foundation in physics first.
And maaaybe our insane priorities involving blowing vast sums on wars and Cold Wars and banning all psychedelic research everywhere in the world half a century ago and allowing GPs and drug companies to get away with literal murder regarding opiates and benzos and so forth might have also had something to do with it...
If Alphafold accomplishes even a tenth of what it promises, they totally deserve a Nobel prize.
I wonder if a similar technology could be used to run prebiotic chemistry simulations? Not in an attempt to replicate certain conditions (like volcanic vents), but just to find out if it could find feasible paths to construct things out of simple organic molecules (like amino acids, nucleotides, and lipids).
This is the stuff that makes Science Fiction a reality. Combine this with Orbital Materials and i get dizzy and need to sit down. I personally can't wait for the mapping the new material 'spaces' and learn insights in reliability, toxicity, environmentally friendly and so on. On the bio side a streamlined series of AI's to detect, design and prepare samples and jump straight into another series of AI's to optimize testing. I bet we can shave off weeks for any need to get something to entire population levels. Exciting times indeed
Finally I will be superhuman
no I'll be first
This aint for us peasants matey
@@cslearn3044 this is for me, Emperor Pigeon
@@cslearn3044 you're right, your 90 year old grandma in the dementia ward will be getting it first decades before you know it she will be evolved far past anything you could ever imagine...
@@grdfhrghrggrtwqqu my grandma is dead
i remember i heard that the protein structures can be seen by very complicated methods but it cost soo much money just to get one protein structure and then see if it can help to certain cases that be able to have a very good approximation using computers is so fucking awesome as it can lead to analyze structures and then see if can be used in certain scenarios without trial and error that cost millions and millions of dollars without even know if will help.
WHAT A TIME TO BE ALIVE!!!
I thought it was spectacular! I wanted to create this article by drawing in 3D in arts with Quest Pro. to publicize. The material is incredible.
Engineered bioweapon dystopia here we come lol
DeepMind's AlphaProteo AI is hailed as a major breakthrough for humanity, offering revolutionary insights into protein structures and biological processes. This advancement promises significant improvements in drug discovery and understanding complex diseases.
Protein folding has mostly been solved? Not sure if there are experts agreeing on this.
I like the future, well, as long as i can afford it.
At least a part of that future will (should) reach most of us for free.
It must be a right...
To health,
To education,
To improve,
With AI.
It must not be only for the rich in developed countries.
*We can't let that happen* .
@@ronilevarez901 How do you plan to prevent it?
@@michaelleue7594 with help.
@@ronilevarez901 yeah a lot of useful free stuff will be there, but a lot of the crazy improvements will cost quite a bit.
@@ronilevarez901 “it must be a right”
Nothing has to be a right, absolutely nothing.
Demis Hasabis is a gift to this world. How wonderful that Google combined with his Deep Mind and continue to help humanity!
WHAT A TIME TO BE ALIVE
So exciting!
even tho i dont understand how do they go from the computer version to actually creating a useful protein in the lab this sounds as groundbreaking as when DNA human sequence was finally "decoded"
what a time to be alive!!!
Alphafold + AlphaProteo is in my opinion Nobel prize material. It's at least as important as CRISPR-Cas9 was.
I think the hardest part is the select one of the many possible bindings in order to minimize risk of side effect, or am i wrong?
To all those hesitant about what 9 to 88% success rate means, it changes your screening effort from thousands or hundreds of thousands of candidates to like 100 and years to a bunch of months, even one month could be enough if you ask. It reduces the discovery cost to like 0.1 to 1% of the original cost. Clinical trials should remain the same, it takes years and are expensive, but manufacturing of proteins instead of fancy small molecules is more scalable and easier to distribuye in CDMO's. As well, protein risk assesment is relatively easier than that of small molecules.
1:03 Alphafold is helping developing enzymes - so is off-brand Tupperware
First thing this will be used for is advanced bioweapons development.
What a time to be alive! 😵💫
Love the alphafolds, alphaproteo 2 will provide tons of vaccines and prophylactics
And tinfoil hats sales will skyrocket.
Love it!
We will soon be having Kryptonions build on our planet itself in our lifetime.
No need for Superman now, we would be self capable 💪
Interesting, but 7 out of 8 tests being between 9-88% successful sounds very much like random results to me. Maybe fairly accurate in specific cases but overall not as impressive as you make it sound.
Do not forget that this can be used in the opposite direction. The moral and moral level of the ruling "elites" makes us think first of all about this and about contacts with military departments.
Amazing!
This is amazing technology ❤
Is the best biography ❤️❤️❤️
i hope we get 1 step closer to solving autoimmune disorders, they sure are a pain to live with!
I love you AI basilisk ❤❤❤❤❤❤❤ what a time to be alive ❤
this is the the basis for biology based nanomanufacturing. plus ecosystem design. Will need ASI for best design.
Been hearing of Deepmind science-domain breakthroughs for five years now and no practical application follow ups. They really need to market practical successes as well to showcase the amazing work
Google don't market anything quickly. Google had LLMs around the same time OpenAI did (see Palm models), but it took OpenAI marketing the technology as ChatGPT for them to wake up and then release Bard/Gemini.
What a time to be alive is something you can say again.
The most important question: ¿could this be used to cure balding?
One can only dream of a new hirsute future for us slapheads
Finally someone brought it up. What is it all worth if we can’t cure baldness?
hope not, i have an amazing head of hair and the baldies need to accept this
In real life proteins have more than one 3D structure. Varies with pH or even electric fields of nearby molecules. I think this is why the practical wet lab experiments generated such a range. It is not the inverse of the protein folding problem,
but the inverse of the protein folding problems. And AlphaFold
before only generated one 3D structure. Also many f proteins combine with metals (zinc fingers, iron in hemoglobin) to be functional. How does this work with non-amino acid componants and resulting shake and charge distribution changes?
Hopefully this will allow me to fix my medical issues ❤
Drug discovery community: "Hold my beer"
Wonderful i would say
And people think Google is behind on AI for some reason. Waymo and DeepMind are the two most successful real world AI projects in existence, both Alphabet companies.
great! finally we can simulate using propability all the possible proteins and what they do! who needs alien worlds proteins now eh? well maybe those never had proteins 😅
This is so huge!
Success between 9% and 88%?
Am I misunderstanding? That's either mostly doesn't work or mostly does work?
What a time to be alive
We will become Elve with this one ❤
There may be a lot of serious diseases that won’t be solved by a single protein or drug, there could be some that will. And it will increase our knowledge base that will help bring some of the more difficult diseases within range of a cure further down the road.
My sister was solving this with the 10 foot phone cord back in 1987
I wonder if and how this will get incorporated into Folding@Home.
Always remember the one true ring can bind us in darkness.
1:40
Including, like, cures for diseases and pretty much all toxins produced by animals/plants (though many of these are more complex than a single protein, it's a big step)
A sufficiently motivated person with bad intentions could use this to make a super bio weapon
Finally a AI model one can appreciate
We’re getting closer and closer guys!!!
The wet lab verification was comparable to current SOTA methods - does this mean it’s not a medical breakthrough yet or does this approach bring major speed or cost improvements over those other best methods?
i think this approach could bring major speed and cost improvement, and reducing the time it needs to verify it, thus this is technically an advancement, they're hyping it because they don't want to look bad to the investor/shareholder.
this approach is more novel i suppose, and it has not been well tested, so we can't currently know if it's reliable or not, i have a fair amount of skepticism here, but the potential impact of this technology is undeniably immense in the long term.
@@CoolIcingcake3467 yeah it seems exciting I’m just trying to calibrate how impactful it will be now versus a couple of papers down the line