Protein folding explained
HTML-код
- Опубликовано: 10 июл 2024
- Join DeepMind Science Engineer Kathryn Tunyasuvunakool to explore the hidden world of proteins.
These tiny molecular machines underpin every biological process in every living thing and each one has a unique 3D shape that determines how it works and what it does.
But figuring out the exact structure of a protein is an expensive and often time-consuming process, meaning we only know the exact 3D structure of a tiny fraction of the 200m proteins known to science.
Being able to accurately predict the shape of proteins could accelerate research in every field of biology. That could lead to important breakthroughs like finding new medicines or finding proteins and enzymes that break down industrial and plastic waste or efficiently capture carbon from the atmosphere.
Join Kathryn as she explains what protein folding is, why it's important and how our Artificial Intelligence system AlphaFold offers a solution to this grand scientific challenge.
Links and further reading:
Find AlphaFold stories at dpmd.ai/471N0Rf and • The story of AlphaFold
Access the AlphaFold database dpmd.ai/474m8Qn Наука
Interesting , but WAY too short, I need a 1h+ lecture/podcast that go into more details please.
I am sure this will come eventually. Right after the Nobel prizes 😁
ruclips.net/video/B9PL__gVxLI/видео.html check out this hour long explanation
☼ they dont have anything to fill it w/. it will just be 1h of claims. and machine learning trained on things we already accidentally know, then more claims.
Lex Fridman has a very good explanation
ruclips.net/video/W7wJDJ56c88/видео.html
@@after_midnight9592 I made a very quick video on the introductions of how proteins fold. Don't know if this is what you are looking for but please check it out :)
ruclips.net/video/WjHmczq3kqw/видео.html
Very exciting. Looking forward to the day Deep Mind earns their first Nobel prize.
Sadly large groups can't win nobel prizes. There's an upper limit of it being shared by 4 people.
@@newtonstan2113 But can't the scientific director or someone have the prize?
@Aayush Sinha AIML Well, this time is computing applied to another field of science, so they should be able to win the Nobel prize of that field.
@@MrPascualex22 Turing prize is the Nobel prize of the computing world
Looking forward to the day they create their own prize lol
Does this help towards development of genetically modified catgirls?
99% of the computing power is working on it son :D
@@carlos84708 We need the full 100%!
@@carlos84708: wait, why 99%? Where does the rest 1% go?
Solar Sands viewer? A man of culture.
@@jerrygreenest 1% trap
Kudos! Outstanding work!
Thanks for your hard work Team DeepMind.
THESE GUYS ARE CRAZY.
deepmind is just insanely ahead than others.
@Daniel G maybe,I mean nobody really knows who is doing what in this field.but they are ahead at least in terms of work process,etc. right??
@Daniel G yes, really. Re: alphafold2
213
2
3
Yeah, but DeepMind didn't quite master Starcraft II. They need to get their priorities in order!
They stopped before they made every pro stop their career haha. Remember, if the bot did not have "human" limits it would own human beings so much easier in games. Human beings never had a chance without handicap.
@@socrates_the_great6209 If the darn thing had to manipulate two robotic arms controlling keyboard and mouse--OK, sure. I could have accepted super-human micro in that case. Still, what I, and I think most others, care about is the level of the strategic thinking.
@@weestro7 Yeah, I hate they had a fixed build order and couldn't apply any reactive tech decisions based on the game's rock-paper-scissors design.
One of the cool things in the original starcraft is that a single zergling will kill a single marine, but a swarm of marines will overwhelm an equal swarm of zerglings, easily due to getting a couple first strikes at range and several kills before the 'lings can engage. Understanding stuff like that is core to understanding Starcraft.
What are you guys talking about, AlphaStar beat the best of the best convincingly.
Stunning accomplishment. Congratulations.!
This along with CRISPR, the future looks very promising. So long as the pharmaceutical and insurance conglomerates don't hoard all the perks of it to themselves.
greatest, beautiful video this year
Beautiful animation 😯
Amazing!
great work guys!! and girls
Yes guys, keep going!!!
Great explainer!
DeepMind, and OpenAI are two of my favourite companies! ❤️ I like Boston Dynamics too. But I'm a bit sceptical about them for their military-grade robots. 🤔
Exciting stuff! :D
So you are saying protein powders in the future will actually make me grow muscle?
impressive !
nice work
What an achievement DeepMind’s AlphaFold Team ! Kudos 🙌🏻
If you tell it the tertiary or quaternary structure you want could it tell you the primary structure?
Interesting content
That last part gave me goosebumps
Brilliant, love the science and the quest for answers. The applications where this knowledge could revolutionise fields from medicine, agriculture, culinary or construction. This seems to be a key step in understanding and manipulating the biggest factor governing biological organisms. The fear in the back of my mind is what a greedy short sighted business could do with this power in the worst case scenario. Off the top of my head, engineer a super human followed closely by engineer a super virus. I'm not sure which would be worse nor what we could do to prevent it from happening, I'm just a guy sitting at home with concerns. If anyone feels like helping a random guy out by telling him these fears won't be realised in his or his children's lifetime I'll be very grateful.
The cells in the body are beyond amazing
How does this compare with Rosetta at home and Folding at home Distributed Computing projects?
Do we know what the protein does once it has folded?
Awesome
I am excited. Looks like a nice progress
I've been waiting for DeepMind to accomplish something outside the constraints of a gaming environment. Well done.
Amazing
Thank you for good work, now onward! first to sleep!
The animation studio who put this together did a great job!
Yeah u us zu
Va ska vi köra ett
@@hxrr1scootz115 jag ska vi köra på det här är det
@@hxrr1scootz115 jag
@@hxrr1scootz115 jag ska du du är
A great achievement!! Look forward to seeing AlphaFold in action
What is the input to the problem exactly? The sequence in which the amino acids appear? How do you know when you have computed a previously unknown shape correctly?
just clickbait.
We're in the future, boys.
We appreciate your compliment..for more guidance WhatsApp.....+::1,,,3,,,,,,1,,,,,,3,,,,,,5,,,,,,3,,,,,,9,,,,,,8,,,,,,.2, ,,,,,7,,,,,,0,,,,,,,,
And girls.. ;)
Lol
@@cryptorevolution9547 go
@@cryptorevolution9547 I okoo
These guys just set an example how collaborative human minds can reach to a groundbreaking extent...
I need science thing to be explained as simply as possible. Am not from a science background so short videos that convey with analogies are good for people with a brain like mine.
Wow!
If my understanding of AlphaFold ll project is correct DNA encodes with an quattuoral language that is three dimensional in nature and the folds are the places where the data sets meet, much like looping and nesting in C++.
Programmers think in bi-dimensional space (think code line numbers) and it seems that DNA encoding operates in tri-dimensional space where not only the line of code but also where in space that line of code interacts with another line of code creates specific protein coding.
Edit: I have read further and also think that gene coding actually operates in a 4 dimensional space with one of the dimensions being the quantum realm where entangled quantum pairs in a non locality pass information between genetic structures.
This means in the future vaccines for Rona-type diseases can be identified quickly !!
I agree, but wouln't it mean people could weaponize protein synthesis more efectively? I mean, the more control we have over chemical/biological processess the greater capacity we have to use them in all their aspects, including malicious ones.
@@noelinsua7261 Everything has its good and bad, be it the internet, technology. To tackle that, we would have to progress on the social front.
@@noelinsua7261 We've had the ability to make devastating biological weapons since the discovery of CRISPR, and yet nothing has happened. You're going to be fine.
They already use computers for that. Thats why moderna started testing their vaccine in February, before the shutdowns
@@TheLegendaryHacker as long as everyone is stupid, you are going to be fine
Fine, but if the body needs a certain protein shape for a certain function, how does it figure out the sequence to do it?
yes
AlphaFold sounds amazing! Loved this animation and video, overall. Hope to learn more about discoveries and progress it has made in the future. I am going to really hope that it can help us tackle things as important as our world's pollution and climate change problems. Without some major progress in those areas, I'm afraid for the future of humanity.
Demis Hassabis needs a Nobel prize
Who did this video? It looks great
Trained on ~100 000 proteins. Seems an impressively small sample to provide such a breakthrough. Would be interested to know about any up-sampling, validation etc.
hmm, how many amino acid are there? some says 20, 21, 22....
The early days of Sony’s PlayStation 3 had what was called Folding@Home. Users could participate in the program and also see a world map to know who else was “folding”. Pretty cool stuff.
This is a big task maybe one of the biggest.
This has more possible variations then the forecasted number of molecules in the universe.
this should have 1 mill likes.
Wow. I'm curious as to how they got here. Really cool to see how A.I. is playing a huge role in social good.
And social bad as well (surveillance systems etc).
@@dislike__button, surveillance systems are used by the police to catch criminals. That's good. Unless of course, you're the criminal.
@Leroy JonesThere are protesters in front of the white house disagreeing with the government. They are not put in prison. Unless you do something criminal, like shoot the president then you go to prison. You have confused the United States with Russia, China, and North Korea.
You're doing exceptional work!
I know it is a long shot, but how exactly does the prediction of "2D shapes to 3D model" help in discovering a cure for various diseases?
By seeing how certain treatment methods interact with proteins, the proteins may fold into a shape that is known to be beneficial, thus reducing the negative effect of some diseases.
once you know the protein shape. what good does it give us though?
protein shapes are the basis of our immune defense
Congratulations to humanity and thanks to DeepMind! ❤️
Will it perfect the Whey Protein to rip my muscles?
This is amazing! Thanks Google. Life comes from movment, the movment of our proteins after an interaction with different signals
what?
And just to complicate things even more, our consciousness decides which signal enter for each of our cells and which signal stay out. When signals enter our cells, than they bind to our proteins to give them their different forms. Really glad Google is mapping all the possible shapes. Powerfull stuff. Hope this power will be used according to their "don't be evil"
This isn’t google
Taught DeepMind was owned by Alphabet but could be wrong. To whom is it?
@@Phyloraptor Yeah Google fund them but they're very independent. They're a British Company called Deepmind located in London
Meanwhile everyone else has no use for AI except for driving up sales and user engagements on their website.
So the shape defines it's function. Mapping out the shapes would be the code to program life.
Once the shape is predicted by the AI, is it possible to verify it using less work than it would take to compute it traditionally?
Yes, in the article by Deep Mind, researchers unable to crack a novel archaea protein by crystallography were able to make light of their findings once DM provided a structure.
They tested it against their data, and found a strong match, whereas the old routes were unable to explain it.
Now we need a paper folding ai
What companies and sector will benefit from this?
Can Alphafold explain how proteins are folded? What’s the mechanism?
yes
I wonder what kind of potential this holds for CRISPR? Can we translate these models to solve for it?
What university field/majors teach this? Is this bioinformatics?
Yeah. Could also be a part of computational biology.
What are others long-standing problems, in any field, that Deepmind could be applied to?
seems limitless to me. the idea
so does this mean the human protenome is soon complete?
What is the margin of error and how is it managed?
yes
@@ishaan863 No
Does this help the fight to cure cancer then?
👏👏👏👏👏👏👏
Great video but choosing to render this in 15fps really takes away from the value :P
game changer in a single generation
Too short, need an alphafold documentary like alphago
Mr and my GYMBROS bout to be on a new type of creatine!!!!!
cool
Guys, I'm a layperson even though pretty curious, how much is it easy for scientists to read a sequence of amino acids of any protein? I mean the step number 1 before you can get started with the structure prediction? Is this task relatively easy nowadays?
Sanger introduced a method for sequencing amino acids out of proteins, in this way you have a chain of amino acids in a sequence linked by peptide bonds ( short peptide) two or more peptides usually interact via bonds disguise bonds to form a secondary structure .This secondary structure can be helix (alpha and beta )or a sheet(beta) they interact to form super secondary structure which in turn makes a tertiary structure (in some cases quaternary structure ) or functional protein .We didn't knew how the structures really interacted at the super secondary or secondary structure to form fold or loops I think they used deepmind to figure that out since it requires a lot of factors to be taken into consideration for proper alignment to build a functional protein
If you know the name of the protein you're interested in, you can simply search it in the many online databases and get the sequence that way. You'd have to be working with a very unique protein (such as from an uncommon bacteria or virus) for the protein sequence to be unknown. The entire genomes of many organisms have already been mapped out - and since protein sequences correspond to gene sequences, protein sequences are known.
Because we have the primary sequences of basically all the human proteins - but we don't the how most of them fold, that's exactly why this is such a big deal.
@@natalyawoop4263 "The entire genomes of many organisms have already been mapped out - and since protein sequences correspond to gene sequences" - exactly answers my question! Thank you so much!
This robot is a genius 😝
The "plastic refuse" that causes marine pollution cannot be easily removed.
This is because they are separated into fine particles as microplastics.
To solve this problem, "Enzyme" that can decompose plastics naturally must be artificially synthesized and then sprayed into the sea.
Yea, but where is Carmen Sandiego?
1:33 "there's lot more work to be done"
HA HA HA 🤣
Внутри каждой клетки вашего тела трудятся миллиарды миниатюрных механизмов. Они переносят кислород в крови, позволяют глазам видеть свет и помогают двигаться вашим мышцам. Эти механизмы называются «белки», и они принимают участие в каждом биологическом процессе во всех живых организмах. Каждый белок имеет определенную трехмерную структуру, которая определяет, что он делает и как он работает. На сегодняшний день нам известно более 200 миллионов белков, и их количество постоянно растет. Однако мы знаем точную трехмерную структуру лишь малой части этого многообразия белков.
Если белок распрямить, он станет похож на нить, на которую нанизаны бусинки-аминокислоты, всего 20 видов. Взаимодействия между этими аминокислотами заставляют белок сворачиваться и принимать конкретную форму из бесконечного многообразия возможностей. В течение десятилетий, сообщество ученых работает над определением формы белка по последовательности его аминокислот. Это важная задача современной биологии. Мы создали систему искусственного интеллекта AlphaFold, чтобы помочь решить эту задачу. Мы обучили систему на 100 тысячах белков с известной трехмерной структурой. Теперь система может точно предсказать форму белка по последовательности аминокислот.
Предсказания AlphaFold поспособствуют прогрессу в самых разных областях. В будущем мы сможем более быстро понимать механизмы новых заболеваний и разрабатывать лекарства для борьбы с ними. Мы сможем использовать энзимы для переработки пластиковых отходов, и даже фиксировать углекислый газ из атмосферы. Все это с помощью белков. Предстоит еще много работы, но распознание формы белков по последовательности аминокислот поможет ученым лучше понимать природу и живую материю.
Nobel worthy.
@0:27 "we know millions of protein but structure of fraction of them are known" Not quite understanding what this means. Knowing protein does not mean knowing its structure as well? Forgot to say, amazing video as always :)
Круто!
As always as Deep mind be the game changer
How much % of world power is from folding@home
Now please try to reduce the neural net to some mathematic formulation (Compression ?)
Finally, a kick to the tick
Letting an AI system generate random probabilities of shapes with the possibility of spitting out the shape folded correctly isn't solving anything.. We don't know why they fold that way, when, or how. This is just allowing us to see what the protein shape would be once folded - but the way it gets there is far from understood. Also, the team themselves have mentioned its only successful on a small subset of proteins. A solution that works for some but not all with no rules isn't a solution either. This is decent for building a company around feeding data into a python application to train, but not exciting at all. We aren't learning anything at all by tweaking weighted values until the AI process manages to spit out a decent answer.
why doesn't adblock work on this one?
cause it is Googles vid and youtube is owned by them, thus making an exceptional ad that passes the usual blocker?
No ads for me.
@@gridcoregilry666 the joke is that the vid is an ad itself
@@goerges388 thought so
wicked innit?
It'd be amazing if the guys of Kurtzgezat - In a nutshell could do a joint video with you guys
Yes, please!
It's just a start
Did it help find a vaccine for COVID-19 or is it bc still in progress?
So as soon as someone makes a better algorithm it’s now Worldwide quarantine
What if we simulate protein using Quantum simulation techniques?
then u make life
There is always a double side to this. Yes you can find dieseases and cures and all that, but the opposite is also true. You can create super diseases as well. And there is always people looking to weaponize things.
This is fucking Dope.
There can be endless possibilities of SciFi disaster films originate from this little clip!
Guys welcome to the future