How AI Could Change Biology
Вставка
- Опубліковано 28 вер 2021
- You've likely been seeing the rise of AI technology everywhere-and some people are pretty concerned about what it could mean for the future. But did you know it might hold the key to understanding, and even changing, parts of our biology? Learn about what AI has changed for the world of proteins in this new episode of SciShow! Hosted by: Hank Green
SciShow is on TikTok! Check us out at / scishow
----------
Support SciShow by becoming a patron on Patreon: / scishow
----------
Huge thanks go to the following Patreon supporters for helping us keep SciShow free for everyone forever:
Chris Peters, Matt Curls, Kevin Bealer, Jeffrey Mckishen, Jacob, Christopher R Boucher, Nazara, charles george, Christoph Schwanke, Ash, Silas Emrys, Eric Jensen, Adam, Brainard, Piya Shedden, Alex Hackman, James Knight, GrowingViolet, Sam Lutfi, Alisa Sherbow, Jason A Saslow, Dr. Melvin Sanicas, Melida Williams
----------
Looking for SciShow elsewhere on the internet?
SciShow Tangents Podcast: www.scishowtangents.org
Facebook: / scishow
Twitter: / scishow
Instagram: / thescishow
----------
Sources:
www.nature.com/articles/s4158...
www.nature.com/articles/d4158...
science.sciencemag.org/conten...
www.sciencemag.org/news/2021/...
www.ipd.uw.edu/2021/07/rosett...
www.thetimes.co.uk/article/de...
www.technologyreview.com/2020...
www.technologyreview.com/2020...
news.cancerresearchuk.org/202...
www.theguardian.com/technolog...
arstechnica.com/science/2021/...
www.vox.com/future-perfect/22...
deepmind.com/blog/article/alp...
www.pnas.org/content/117/3/1496
emerj.com/ai-sector-overviews...
www.chemistryworld.com/opinio...
www.dw.com/en/what-are-protei...
www.nature.com/articles/d4158...
www.nature.com/articles/s4158...
www.nature.com/articles/d4158...
www.theguardian.com/technolog...
www.nature.com/articles/d4158...
www.imgt.org/IMGTeducation/Aid...
predictioncenter.org/casp14/d...
• AlphaFold: The making ...
deepmind.com/blog/article/Alp...
deepmind.com/blog/article/alp...
www.nobelprize.org/prizes/che...
www.frontiersin.org/articles/...
www.ncbi.nlm.nih.gov/pmc/arti...
• Understanding Crystall...
Images:
commons.wikimedia.org/wiki/Fi...
commons.wikimedia.org/w/index...
wellcomecollection.org/works/...
commons.wikimedia.org/wiki/Fi...
commons.wikimedia.org/wiki/Fi...
commons.wikimedia.org/wiki/Fi...
www.eurekalert.org/multimedia...
commons.wikimedia.org/wiki/Fi...
commons.wikimedia.org/wiki/Fi...
commons.wikimedia.org/wiki/Fi...
commons.wikimedia.org/wiki/Fi...
commons.wikimedia.org/wiki/Fi...
commons.wikimedia.org/wiki/Fi...
commons.wikimedia.org/wiki/Fi...
commons.wikimedia.org/wiki/Fi...
Hank missed a golden opportunity to say "the future of biology is unfolding."
in biology, you usually don't want stuff unfolding, hank didn't 'miss' an opportunity, you missed the point of the video, and an understanding of complexity.
Don't worry though, you & I are both just silly lumps of soggy cabon zooming around on a tiny space rock that orbits a tiny star. \o/
Yay!
Or rather, folding.
He said 'take shape' which is closer to the subject at hand. 'Unfolding' is the opposite. No need to thank me for ruining your pun :D
@@Miranox2 it's all occurring in a lab anyway, so you might say it's de-natured..
@@TheOrganicartist Yes, but in writing, "unfolding before our very eyes" is a poetic way to describe the occurrence of an event. Jack was correctly alluding to that bit of wordplay.
As a molecular biologist myself, I can see why this makes Hank present this episode with and extra dose of passion. This is a revolution that shortens years-worth of lab research to days or even hours.
Same... A few weeks back I just wanted to compare a protein I'm currently working on to some hit I found in the database and I was really surprised, when I saw on uniprot that the 3D structure was predicted by alphafold. I'm so glad I'm only 25yo in this day and age, where I can see this dream become a reality.
In before someone from Facebook decries it for being unresearched/rushed
as a non-molecular biologist, this is neat and i hope it makes science even sciencier
I have a question. Since this program can predict how a protein would fold and the amino acids used to make it. Let's say you designed an awesome protein and wanted to make it in the real world, how would you go about making it?
@@dondai2880 Let a DNA or mRNA make it for you.
I'm starting my PhD studies in a few weeks, the subject being protein folding (with NMR). Pretty excited about all this AI stuff !
Any updates??
Give us an update buddy
I wish my biology teachers at school had given half a hint at how exciting and amazing biology is! Hats off to the teams and all the biologists who made this possible!
You must understand that 99% of people do not understand the implications of technology. Most will disagree even when you present them the evidence.
I love the scishow episodes where you can tell Hank is actually excited.
You mean... All of them?
I understood not much of it but damn Hank was excited today
@@jimbrookhyser he’s always excited but today he was extra excited and it was amazing!
That's the core competitive advantage of SciShow, isn't it? Hank Green enthusiasming all over the screen?
It's a beautiful thing
The fact that discoveries like this are made free for academic use gives me not only hope in humanity, but hope FOR humanity. Thank you to everyone involved in this research.
Deepmind was kind of pressured to do it, they would have monetized it otherwise
@@tresuvesdobles good that they were pressured, then
Mhm!
@@tresuvesdobles "haha look I am shouting at a corporation even when they do cool stuff, I am so cool"
@@GMPranav Such insight, comments like yours definitely make yt a much better place
I remember a few years ago, making a prediction of a protein's structure from it's sequence even by seq comparison was impresice and time consuming, this is truly revolutionary
I love how excited Hank is about this. Amazing stuff! Will be cool to see where this goes.
I'm a biotech engineering student, and this just blew my mind wide open. I cannot fathom how much the world is about to change. These kinds of dicoveries and inventions are keystones in human evolution, and I'm happy just to live in the same time as one is being discovered/invented:). It would be dope to apply it in a couple of years in my career and bring something good to humankind, though
I love it that they made it publicly available almost immediately, as if it was the only right thing that can be done. This is the best of us.
We look forward to your ground breaking biology paper enabled by proteinfolding AI :)
I dont know anything about the science behind this, but this video makes it sound as revolutionary as the mastery of Fire. It sounds like we could develop a spray which would turn an Oil Slick into nutrients for Algae that settles to the bottom of the water column. Things like this in the next decade would change the nature of our world.
Eternal life here we come
@@IHateUniqueUsernames there will always be good people who look for the betterment of the world and care not for money or fame :)
I can only wish to be 1% as good and kind as them
As awesome as this is, the coolest part for me is how they immediately shared the full methods and codes 😎
Yea for China and the CCP. I am sure they will put it to good use and weaponize it, and of course file a patent and claim it as all their invention.
@@kevan5321 sharing with everyone immediately prevents that or a corporation from doing the same.
@@kevan5321 also CCP have no ethicals, they will research anything for any cost.
That was very cash money of them
after all it wasn't from apple company 😅
So happy both groups released their code - I hope important tools like these always remain free and open to use!
The future of biology is unfolding before our very eyes 🧬
Suddenly it feels like I might be able to see cancer become a curable problem in my lifetime. Just imagining people getting cured from cancer without going through the horrors of chemo is so heartwarming.
Even better, this might be able to allow the human body to process sugar and carbohydrates like dietary fat and prevent cancer, obesity, and diabetes while allowing people to eat whatever they want.
My goodness.. how is it people still wallow in darkness... cancer has been curable since the 1930s . Royal Rife is what u want to input
@@matthew8153 I'd run off chocolate milkshake and doughnuts they my fave
Dear. There is a clear treatment for malaria. Yet millions die per year. It is not only about science.
"people"
Yeah right
Ultra rich elites
The way he presents we can see how excited he is about it. He made me excited too like I am gonna do something with it tomorrow even though I don't have any science background at all. 🤣🤣🤣
Yeah, he's so much more energetic today.
He has the excitement of someone who's had to do protein crystallography before
I am studying Biology in the english Curriculum for A-Levels. I am excited to hop into this world after a few years. I can't wait!
Being able to simulate protein interactions is going to be absolutely massive. In the future, being able to safely predict what will happen when a gene is edited with CRISPR will permanently alter the course of human evolution.
That is insane to think about!
terrifying. who can guarantee this power will be used wisely.
@@heinzie5 there will always be those who misuse technology(nuclear science lead to nukes), it is all but a tool but on the upside being able to treat genetic diseases, finish of cancer and other stuff is important on levels i cant describe.
@@heinzie5 It's obvious that this power will be used both wisely and unwisely anything else wouldn't be human!
we're gonna need a more heat tolerant and drought tolerant version of ourselves pretty soon.
This is the exact meaning of “AI is a tool.” It can’t do everything, but it can be *used* to do everything.
This feels like one of those things that is a flashback in a history lesson of some utopian society hundreds of years in the future.
"In 2021, two teams of researchers made a breakthrough understanding in our ability to simulate protein folding, and then released their techniques freely for academic use. This ushered in an unprecedented era of human progress, giving us headway towards final victory into some of humanity's largest problems which had been seen as practically intractable up to that point."
Flash forward to present day and play out the story describing all the ways it fundamentally improved the world, and the ways those developments played out on a societal level
Making me cry
@HoboGardenerBen yup, most inventions nowadays in society are solutions for problems that previous inventions created. And one can assume that even inventions like this will wind up with consequences that'll continue the spiral
Regardless, it's fun to speculate
@WalkAMileAncientCityStyle That's an extremely dangerous mentality
@HoboGardenerBen Lol no system will ever be able to make a utopia simply because one mans utopia can be another persons hell. But capitalism + democracy is by far the best system have got. All other systems have either failed or made less desirable results.
But problem with capitalism is you need to have balancing forces to stop runaway wealth accumulation so few people can control the market and also use their wealth to control politics/government, which why you need to have strong antitrust laws, bribery laws ect. and enforce them.
This comment reminds me of
m.ua-cam.com/video/CWu29PRCUvQ/v-deo.html
m.ua-cam.com/video/-IgMUPZ_Ekc/v-deo.html
The protein structures finally became so complex, they were able to analyze themselves and start moving towards more intelligent system design.
I assume you are talking about how we're made of proteins and are analyzing ourselves
@@Emmet_v15I would assume that you assumed correctly
Well actually 😂
@@daddymememaster5432i would assume that you correctly assumed that they assumed correctly.
Definition of biochemist:
DNA with knowledge of self.
I’ve worked in the crystallography field since starting grad school back in 2003 - this is an absolute revolution. Without even looking it up, as soon as I heard Univ of Washington, I knew David Baker’s lab would be involved with this somehow. I love reading their papers.
I'm terrified by the toxicological implications of this. It took us decades to realize that many plastics had long-term, non-acute interactions. Being able to make a custom protein that does a designed task wonderfully is good news -- but what about the Law of Unintended Consequences for these designer proteins? I hope like hell the protein designers consider toxicology tests before racing to wide-spread usage.
This is such a massive step forward in biochemistry that I can't even predict all of the applications it will bring forward.
Maybe we should design an algorithm that can!
well, that's exactly what scientist thought about nuclear energy and back them no one predicted it will be misused for killing thousands
@@great__success the application of nuclear sciences for all purposes, including strictly military ones , have helped more people than it has harmed
@@crovax1375 l
@@great__success The inventors of the hammer probably never intended for it to be an improvised murder weapon tens of thousands of times but without it could you imagine how many houses wouldn't have been built?
this is the best news ever. my bio professor was talking about how we always need to verify that the protein simulated matched the real life one, but considering the AI’s high rate of success we essentially eliminated a ton of the work to get there. here’s to a very productive and prosperous decade of biotech!! 🥂
Lets not do anything messed up with it like alot of world changing tech.
@@NineSeptims Like any tool or technology, it will be used for all purposes. And the field of molecular biology and protein design is going to provide a large number of incredibly powerful tools.
Within 100 years (honestly, at the rate this is going, maybe within 40) we will be able to cure any disease.
I'm not joking. We will be able to cure any disease. This is the field that will do it.
If you want, I'll type up a whole damn essay about how cool this stuff is. I actually worked at the seattle lab mentioned in this video for a semester in college.
But, it's also not hard to come up with a way to use this technology for evil. If someone was inclined to use this technology to develop a bioweapon --- hooo boy, that scares me more than nukes.
Cure every disease. Wipe out humanity. It's up to us what we do with it.
Frankly, I'm expecting both.
Knowing biologists, we will probably have to still confirm the protein structure. Similair to how we do an in silico analysis (computer analysis), and then have to do an in vitro analysis (lab analysis) after that.
omg this is great. I am getting an MS in Bioinformatics so this is up my ally. I used some protein predicting tools and it has been so confusing since the results often were different in protein families and the protein family functions. I am very excited to be able to use this.
I absolutely love your exploding enthusiasm! I can see this topic is one you really care about. Thank you, fascinating and clearly explained - as usual.
I'm a bioinformatician, this breakthrough has been heard since June or July this year. It is blowing up my mind how advance our technology has come. Luckily around the time, the software was already open to be used and I was lucky to be able to use it for my reasearch.
Does it reay huge advantaged or just overhyped?
Yes. please tell us how large an impact it's had on your processes and result times.
In my case, I was able to get a better 3D structure model for my enzyme (lipase from a specific bacteria) and with that, I did further analysis to introduce a mutation in the lipase to alter its hydrolysis behavior (more than one program were used).
I also compared the bioinformatics analysis results with the lab experience.
The bioinformatic has predicted the hydrolysis activity increased and it's also proven by the lab experience
So ya, it has a huge advantage for me cause it can "predict" the enzyme before the lab experience
@@djamil59ify
The hype was 15 and 20 years ago. The technology is now a reality and is being applied to the real world.
The reason this has happened is due the hardware engineering and software engineering and ai research.
It is the perfect meeting point of different capabilities.
@@vincentchen9678 that's insane! so, you can predict what happens under specific circumstances? do the people working against the disease of the day use this to sorta see the effects of a drug? i am so curious now i know it works and is very useful!
As a structure biologist this is such an exciting time to be working in this field! Also love the palpable excitement from Hank!
Was a bit sad though, that cryo electron microscopy (cryo-EM) didnt get mentioned as one of the methods used to solve protein structures (maybe I am biased since thats the method we are using in our lab).
In cryoEM we flash freeze proteins in liquid ethane (-175°C) which preserves them in a near native state and then image them under an electron microscope. We collect tens of thousands of images of these frozen "in time" protein particles and then use complex algorithms to generate a 3D representation of the protein (also referred to as map).
as a molecular biologist i was also expecting him to mention cryo EM
How many times in a day can you do that?
Cool- whats the rough timeframe for bulding a map for a new protein?
What are main limitations- eg are certain proteins outside the scope of C-EM?
Thanks!
i'd imagine TITN (Cant remember the name accurately), the 180,000 character protein. That's a lot of computations of all of the atomic & quantum effects, bonds, electrons, etc. But that's what supercomputers are for.
As a biomedical engineer, I was also expecting him to mention how the crystal state not necessarily how the protein is folded under normal conditions. CryoEM I believe can get a more complete picture of the protein in all its conformational states. Interested in if I got that right, from the structural biologist haha.
So fun to see you so excited Hank! Loved this topic!
Thank you for helping me understand this. Translating these complicated concepts into layperson-speak must be really difficult. But the effort is truly appreciated!
Let's all geek out together with Hank! This is biology's and medicine's singularity.
I've geeked out so hard I've transcended Time and Space. Send help.
@@procrastinator99 sending an Uber. what's the address of that continuum?
It's only the beginning. Give it a few more years for the baby universe created to truly inflate
Good timing...was just researching protein folding last evening.
prions?
@@DyslexicMitochondria Hey bro i watch ur videos. Love ur channeI
Similarly, my protein folding consisted of bending my hamburger patty to fit a hot dog bun.
@@DyslexicMitochondria Now I need to go find out about prions. Ha.
So... prions can cause Alzheimer's as well as other terrible diseases.
Prion replication is still being studied. Prions junk up the extracellular matrix which seem to be toxic to neurons.
Don't want this being weaponized into something easily contagious.
This is awesome and love to see your excitement on the subject. Can't wait for the great outcomes of this technology.
Since I'm now in my 50es and several of my close relatives have died of cancer, I'm naturally very interested in any news about potential breakthroughs in cancer research. For years I have assumed that I'm very likely to some day die from cancer, but recently I've started to doubt if that is necessarily going to be the truth.
Your excitement is infectious, Hank. This makes me so damn excited about the future of medicine.
This may be one of favourite ever episodes of SciShow. Hank is so enthusiastic that he looks like he’s in a VlogBrothers video! This is seriously, seriously cool and I can’t wait to see what humans can do together with this new knowledge!
This is amaizing. Probably we cant even imagine the possibilities this kinds of technologies will bring.
We used alpha fold in my biochem class and my prof was trying to explain how exciting this was, but i didn’t get it until now. Thanks Hank!!
I am a PhD student who worked very closely to this field of research. This is very understandable but impressively accurate! Keep up the good work!
This was so good, the way you’re able to break down such detailed topics so concisely is incredible
This is really something else. This is so significant.
Thanks for putting it so clearly, Hank!
Ahh! Exciting video and great covering of the good news! I work here at UW in the Institute for Protein Design on new peptide-based therapeutics so this subject is very near and dear to my heart. Us graduate students have felt the impact of this development and many of us are eagerly using it to further advance rationale drug design and medicine!
I really don't think I've seen Hank this excited before
Sometimes, when I watch shows like this, about how humans are advancing science. I start doing a little dance in my seat. Just me? This stuff just makes me happy, lol
Yeah it really is amazing!
Not just you :-)
its you being a proper human
Very exciting! Thanks for making this so easy to understand. Huge!!
This is exciting! Your narration was perfectly concise and inviting of creativity. Cheers!
This is incredible!
When are people going to wake up and say NO you are not putting anything in my body even if your promise me ill live to 2pp years and will not need health care! Time to wake up people Jesus came back and paid the price for Fully Human not hybrids. There is nothing knew under the sun this genetic alterations went on in the times of Noah. It says the Nations will be deceived by sorcery which is the word Pharmacy!!!!
@@travismckinnon2494 I bet you still use tylenol, advil, tums, pepto, and go to the doctor when you're sick. If you don't, kindly go live in a third world country. Stop trying to turn ours into one.
@@natalyrausch I don't use those products thanks to God. He showed me what they have been doing to our food and water so taking care of that problem will cure 80 percent of people. You think advancement in gene editing is good and new but its bad and old. It was done many years ago what do you think dinosaurs were genetic experiments hence why none are alive today.
I can’t wait to see how AI helps humanity in the future.
@@johnconnor7501 Had an ex a project manager at the I won't say the name for her protection but it was at one of the largest tech giants you can think of 10 years ago it was far beyond Skynet back then. What they show you now is nothing. Not that many months left to go until Judgement Day.
As someone who wants to go into biology, this is so cool and exciting!!! Thanks for spreading the information :)
You say biology, I say wild nanotechnology. :p The potential applications of this are endless.
The complement system is a good example of the power of proteins; Kurzgesagt has a fun video about it; ua-cam.com/video/BSypUV6QUNw/v-deo.html
Not everybody is cut out to go into biology or sciences. You need to be sure that you have the natural ability to perform in such fields.
You may find that you are better suited to other fields or that biology is perfect for you.
Such a pleasure to see you just...so VERY enthused.
What a curious discovery, I had no idea that this problem existed let alone that there was a solution to it! Thank you for the explanation, I've learned a lot and I might even check some more stuff about these problems at a later stage! You guys and the world have never cesed to amaze me.
This is incredible. One thing that immediately comes to mind: since using a classical deterministic algorithm protein folding is NP-complete, and all NP-complete problems are computationally equivalent, does that mean that we could use a similar approach to solve/approximate other NP-complete problems? That would be incredible
I wish I was smart enough to understand anything you said 😂
NP complete stands for nondeterministic polynomial-time complete. My question is how this relates to topology, the mathematical theory of knots.
Technically this AI is not a "solution" since it doesn't give the right answer all the time, more like an heuristic comparing with NP problems. We've got some good heuristics for most NP-complete problems, it really isn't that worthy. I work with AI and deep learning, everything from feature extraction to training it's very time and resource intensive also engineering an ensemble neural network with hundreds of layers is very hard and costly to optimize.
While comparing with NP problems this algorithm might be seen as a stochastic output tending to infinity (depending on the training) so id consider it to be Omega(infinity) aka not a solution for NP problems.
Doing something actually useful with machine learning costs a lot of money, you need someone to give you those resources but it's very fun and powerful lol
For anyone reading these comments, 👆👆👆 this comment is the best answer
There are algorithms that give you good enough solution to NP complete problems. This AI is in similar category it gives you a good enough prediction of the protein shape. But it does not give you a mathematically correct answer only answers that are close enough.
Machine Learning is a crazy powerful tool! I'm so excited for the future it's bringing
im not. it'll learn were not worth saving.
@@TheOppiter negative nancy
Just beautiful. I have actually sat down and attempted to work out protein folding on paper just to gain a better understanding of how this occurs in nature. This is fascinating stuff. Wow!
Every once in a while I watch a video that I wish I could give a "Super Like" to or something. Watching Hank so genuinely excited made me grin ear-to-ear for almost the whole length of the video. This was so good
Your enthusiast is contagious; I now love protein-folding-AIs too. Thank you as always for the wonderful content!
So exciting, between CRISPR and this, biology has broad frontiers opened up for generations to come.
Don't forget the recent addition "CRISPRoff". I think we will see it soon (in relative terms) in medical use.
@@shockcat5988 that could be said of many scientific breakthroughs. But still it's much better for this to be open to public rather than be secret knowledge of some country's army, right?
@@shockcat5988 The same technology can be used to fix all those problems (at least eventually) you mentioned.
Or .. Create Nightmares...
@@shockcat5988 Technology is a tool. It is our responsibility on what we use it for
As a Biochemist, this technology can be applied to so many areas! I'm excited to see the change that this will cause. Hopefully, it will be used for good!
This is truly a revolutionary breakthrough! Protein folding was notorious as the utter and ultimate barrier to understanding all of life, and it was so difficult that many saw it as marking the absolute limit to human knowledge: just too difficult & complex to ever be solved. Let's just hope we treat it with the awe and respect it deserves. We already know of diseases caused by protein mis-folding...
Yeah, only issue ive got is that humans don't always use these things well. I think we're developing a bit too fast and nobody is bothering to slow down and think about the bigger picture outside of "what can this do for US?". Approaching all of these new frontiers, with some people saying that it's bordering mystic understandings of things...It's potentially dangerous when we start messing with forces beyond our control or ones that we might not be ready to handle. (We thought the atomic bomb might theoretically blow up the world, and yet we still set it off for the sake of being able to destroy others). Just makes me slightly uneasy how rapid everyone wants things now.
@@zaclovesschool2273You can, perhaps, delay an avalanche. But you can't stop it.
This...this is like a baby titan starting to stand on the shoulder of giants!
Hank's enthusiasm was truly infectious.
This is so amazing. Cant wait to see if this can be applied to my studies!
I played this vid as a bg audio while I work.
But then, I can hear out-of-usual excited voice nuances from Hank and some keywords that make me "Wait... What?"
So I focused on it and put it on full display on the main monitor.
It really is exciting stuff!
Man, what incredible news. The potential this breakthrough has is absolutely unending. I mean, the sheer diversity of structures made in nature by proteins along is immense. Who knows what developments well be able to achieve with this new method? I cant wait to see what comes out of this
*"Who knows what developments well be able to achieve with this new method?"*
Dont worry, the rich will find a way to get richer with it by making our lives more miserable
@@faustinpippin9208 quit being so negative
@@babecat2000 In this world? nah
I was wondering where sci show was this eve... This explains it, this is huge!
I remember as a biochemistry student in the nineties dreaming I’d some day come up with an insight that would allow me to predict protein folding using an Excel spreadsheet. I seriously overestimated Excel and underestimated my own stupidity. (Although one of my fellow honours student though she might have cured cancer using peptides).
Love the content! Love the shirt combo, Hank!
These videos make my day and especially on topics that involve the health of others. Almost brought a tear to my eye and I'm extremely excited to see how this develops.
The excitement in Hank's body-language is unambiguous in this ep :)
Fascinating! Hoping future versions of these models can incorporate how the composition of the phospholipid bilayer impacts on the structure and folding of membrane-bound proteins.
Amazing video !!!! Really appreciate!!
Considering my chronic illness deals largely in errant proteins, I am thrilled to see any progress in this area of knowledge but this is beyond! I hope this will be even more helpful to scientists and doctors than expected. WOO!
super cool stuff! I remember running Folding@Home a long time ago which was trying to do just this
You can really tell how excited Hank is while making this video, I'm so happy to see that
Thanks for excellent research and presentation
I’ve been waiting for someone to make progress in the protein folding conundrum for many years now. And the fact that it’s now happening makes my day. Thanks for the update!
ahhh, the leap forward everyone has been waiting for......will be EXTREMELY interesting to see what they do with it! Can't wait and see =)
Hey hank! Just wanted to say thanks for everything you do and continue to do. I'm 25 years old, and I can't count the number of things I've learned from you. It's kinda weird to think about, but you've been educating people in science for a huge portion of my life. I remember teachers playing your crash course videos back when I was in middle/high school. You are clearly passionate about what you do, and I just wanted to acknowledge your hard work. You've taught me more than most of the teachers I've had in-person, so thank you
The math is wrong, laughing out loud. Lol
Discover by yourself how to do discrete math using algebraic methodology. You were born for calculus. I won't calculate for you. The humans left books for that matters.
My grammar is still incorrect, like 125.
Nice video, thank you. It would be nice to see more similar videos
I can just feel Hank's excitement as he is describing all of this :)
This is genuinely amazing. Im glad I got to see this and how itll change our understanding of the world in the future
Thanks you sir, Your analysis on AI inspired me and deepened my interest on computational biology. As bio postgraduate with computation interest, it is fascinating what can be done with both combined. It would be great if you could make an analysis video current scenario of this algorithms after the current advancement in AI field like OpenAI and other. It could inspire generations of new age biologists. Brief about career on this field. What are the roles one can expect. Your videos are always inspiring. Thank you again.
Great video, thanks. Would have been even better if there was mention of the RNA folding prediction work of Raphael Townshend and Prof. Dror at Stanford that made it onto the cover of the journal Science in August. As far as I understand it, they now have the bragging rights to the most predictive AI algorithm.
That sounds revolutionary. Also, great video!
So wonderful to see Hank so, so, SO excited :D
His energy in this is better than a cup of coffee for my brain haha
Just don't drink coffee when you're stressed out, it has plenty of side effects.
Scientist here.... with a humble request ... can we PLEASE put the breaks on any novel prion designs without a LOT of explicit oversight? Its a small thing to ask, really.
Loved the in-depth explanation!
I do research using alphafold and its ridiculously accurate. What people aren't talking about as much with alphafold is that it can also accurately predict the structure of protein complexes reasonably well.
Edit: It seems much more accurate than rosettafold as well
But he talks about that at about the 4:30 mark though?
Their predictive power is still limited to the training datasets, thought, right? Like, there may be types of folding it's never been shown. And (systematic) errors in the experimental data?
And isn't protein folding essentially a computationally irreducible physics problem? Like, you can get close a lot of the time, but there's no shortcut to 100% accuracy. I thought I'd seen some more negative headlines to this effect, on this story, recently...?
@@Krakenman123to my knowledge alphafold isnt designed for complexes but it still works (you can trick it by connecting sequences with flexible amino acids which acts like a tether)
@@ZeroGravitas we have experimentally validated so many protein folds up to this point that we have found all major types of folds, like helices, sheets, turns and unstructered regions. To find another completely new basic fold would be like reading novels in the every language on earth for decades and then suddenly finding a new punctuation mark that's never been used before. Not one somebody created by purpose, but just some new kind of comma or exclamation point that was always there, just never used.
Sure, it's possible, just nearly impossible.
And to the systematic errors in the experiments: could be, but then that error should also inhibit the development of new drugs, because they're designed according to specific protein shapes and they work, so there can't be a major error in our methodology.
And lastly: if we figured out the rules to protein folding, like we did with the smaller structures, like helices, we could predict them accurately. But the rules are so complex, because every amino acid can influence every other amino acid, depending on distance, the other AAs in between, the pH or polarity of the surrounding environment... That we just can't disentangle them well enough. So we're training a NN to get an intuitive understanding of the rules. And it's accurate enough. 100% can't be reached, since we're talking about atoms and they always vibrate and move in their bonds, etc. But we're already as close as we can physically get, so the problem isn't actually one.
It's really a massive step forward towards future biology.
I can imagine that there is a tremendous potential of applications this technology will provide
Being in my 40s, I see hope to live longer...
Congratulations to everyone related to the research and achievement...
Mankind will remember you..
It may not be purely altruistic, they could be intentionally or unintentionally helping themselves (or a mix) to live longer. Nothing wrong with that if we can somehow keep population and resources in control. There's a lot of other comments with negative viewpoints (use as a weapon, this is against God and religion, corporations will hoard it.. etc etc.), I'm just stating that there's definitely pluses and minuses that now start to pop-up that hadn't been there before. I'm excited as anyone who's in their latter years in hopes to prolong a healthy life with technology; I don't want it to cost the future of humanity though.
@@sirsia1st
They?
The research algorithms are completely open source and published. No one company or person will control this knowledge.
@@bighands69 I certainly hope you're right, but my life experience is different. The rich and mighty keep getting rich and mighty, while the rest of us live with the scraps that fall off the table's edge.
@@sirsia1st
Once something has been achieved everybody will be able to figure out how to do it. That is the beauty of a market economy. Competition creates lots of choice and variety.
@@bighands69 optimistic. Who here remembers the first scientist. To discover methane eating bacteria with a high growth rate that were also high in protein.? What were those scientist names now and when was their discovery made ?1940s?. And what. Have we done with that incredible information?
Woo! Great timing on these AI discoveries and assists - good training on those
I collaborate with an x ray crystallographer who has started running the amino acid sequences through alpha fold before he does the crystallography and he said it's so accurate he's worried he's going to be out of a job pretty soon. It's wild that something so powerful is free to the public and can be run on a small server.
How much of crystallography is in silico these days...?
I was wondering when they would cover this , even as a layman my mind was blown when i read an article about this. I share Hank's excitement!
Hank I've been watching these videos since 2015, I graduate this summer and once I have a cool science job, I will sign up for the patreon. I love what you guys do here!
This is so exciting!! Thank you Hank, that was a wonderful video.
Wonderful! Love your enthusiasm
I was just gonna say, I've never seen Hank so excited before!
every biochemist: "great, we know what it looks like... BUT WTF DOES IT DO!?"
Structure is an excellent and vital stepping off point, but understanding the function and how the structure informs that- that is the key
Well, the point of this isn't that the structure tells us what it does, the point is we can design and create artificial proteins much faster, which people will than use to test the effects of in much shorter times. Essentially this speeds up the process like, 100's of times over
prediction of accurate structure could allow accurate bioinformatics prediction such as active/ cataytic site determination, docking prediction and MD. Also, there are many publicly available bioinformatic tools for protein analysis, most of them require 3d structure. So basically knowing the structure is the first steps for protein revolution
been working on this and I must say it's really good
I think some people are underestimating how truly amazing this is. This could mean solving basically almost every disease, as well as aging.
They really are. Especially now that LLMs have gotten so much attention. (Although the two are related, of course, like all AI.)
AlphaFold came out, and then two years later, AlphaFold 2, a substantial improvement. And then more. And this is just the beginning. This March, a drug that could treat liver cancer was designed in 30 days by Insilico Medicine. What the hell will they be able to do in ten years, with vastly greater computing power?
I used to be quite conservative about aging science, Aubrey de Grey, David Sinclair etc. I figured it would take 20 years to produce something viable. Or longer. Aging is very complex and the funding was not there. But my views have changed as these technologies have appeared and money has poured into longevity research. The least the rich assholes with giant yachts can do for us is fund this research :P
It's been a year, did it change anything so far?
@@adryncharn1910 Research and implementation takes time and also money. You have to do experimentation before being able to accurately tell if a protein works as intended.
I'm sure biochemists are already working on it right now, it's only a question of time.
"The future of biology is starting to take shape" - this must have been the most subtle pun you ever made. Good one.
(for anyone that didn't get it, hint: protein folding, taking shape, position)
This is going to be huge, like sci fiction big. same thing with fusion, where its impact is just so wide and far on what we can and cannot do as a society.
Going to be following this closely.
Exciting times!
Fusion is much less close and tangible than this.
Fusion is not yet realistic and is akin to the ai promise of 25 years ago.
Another fabulous episode making complex scientific concepts clear to everyone! Thank you!
Neat, now imagine if we could throw it into reverse. Create a protein model with a novel shape, describe the properties you want specific regions to have, and feed it into software that analyzes it and tells what sequence of amino acids are needed to create a protein that will fold into the structure you designed.