AI Firm DeepMind Puts Database of the Building Blocks of Life Online (theguardian.com) 19
Last year the artificial intelligence group DeepMind cracked a mystery that has flummoxed scientists for decades: stripping bare the structure of proteins, the building blocks of life. Now, having amassed a database of nearly all human protein structures, the company is making the resource available online free for researchers to use. From a report: The key to understanding our basic biological machinery is its architecture. The chains of amino acids that comprise proteins twist and turn to make the most confounding of 3D shapes. It is this elaborate form that explains protein function; from enzymes that are crucial to metabolism to antibodies that fight infectious attacks. Despite years of onerous and expensive lab work that began in the 1950s, scientists have only decoded the structure of a fraction of human proteins.
DeepMind's AI program, AlphaFold, has predicted the structure of nearly all 20,000 proteins expressed by humans. In an independent benchmark test that compared predictions to known structures, the system was able to predict the shape of a protein to a good standard 95% of time. DeepMind, which has partnered with the European Molecular Biology Laboratory's European Bioinformatics Institute (EMBL-EBI), hopes the database will help researchers to analyse how life works at an atomic scale by unpacking the apparatus that drives some diseases, make strides in the field of personalised medicine, create more nutritious crops and develop "green enzymes" that can break down plastic.
DeepMind's AI program, AlphaFold, has predicted the structure of nearly all 20,000 proteins expressed by humans. In an independent benchmark test that compared predictions to known structures, the system was able to predict the shape of a protein to a good standard 95% of time. DeepMind, which has partnered with the European Molecular Biology Laboratory's European Bioinformatics Institute (EMBL-EBI), hopes the database will help researchers to analyse how life works at an atomic scale by unpacking the apparatus that drives some diseases, make strides in the field of personalised medicine, create more nutritious crops and develop "green enzymes" that can break down plastic.
Re: (Score:1)
My ad blocker didn't block that either. WTF
Also consider kids may be viewing this. WTF
And dromeda (Score:3)
and develop "green enzymes" that can break down plastic.
Wasn't that the big bad in The Andromeda Strain?
Re: (Score:1)
Re: (Score:2)
Hey at least if shit hits the fan, it will only break down plastic and not our lungs... For now.
Re: (Score:2)
at least if shit hits the fan, it will only break down plastic and not our lungs...
And if you happen to have plastic heart valves, or a plastic stent or a polyethylene hip replacement, it's just bad luck...
Re: (Score:2)
Your PC's and phones will get the equivalent of termites.
Right on! (Score:2)
AI Firm DeepMind Puts Database of the Building Blocks of Life Online
This is ironic given online info is regularly used to initiate the disgorgement of the building blocks of life.
Off the beaten path. (Score:2)
DeepMind, which has partnered with the European Molecular Biology Laboratory's European Bioinformatics Institute (EMBL-EBI), hopes the database will help researchers to analyse how life works at an atomic scale by unpacking the apparatus that drives some diseases, make strides in the field of personalised medicine, create more nutritious crops and develop "green enzymes" that can break down plastic.
And help create chimeras.
And the reason for this PR piece... (Score:2, Offtopic)
Computer scientists are questioning whether Alphabet’s DeepMind will ever make A.I. more human-like https://www.cnbc.com/2021/06/1... [cnbc.com]
DeepMind reportedly lost a yearslong bid to win more independence from Google https://www.theverge.com/2021/... [theverge.com]
Trying to justify its existence in the free market....
Re: And the reason for this PR piece... (Score:3)
PR piece or not, profitable or not, this is useful to biologists and scientists who care about the proteome. It provides a lot of leads for proteins that are not well studied. There are tens of thousands of human proteins, but the function of maybe only one thousand are studied thoroughly. When applying for a grant, they will not care much about your study if it is not studying be one of the cool genes. There are thousands of proteins that have only 3 or 4 papers published about them, without even structura
Re: And the reason for this PR piece... (Score:4, Interesting)
Knowing the structure of a protein is very useful in figuring out what the protein does.
To be more specific, it's almost impossible to understand what a protein does if you don't know the 3D structure. For example, some bacteria use a protein shaped like a hollow needle to inject things into human cells. If you don't know that it's shaped like a hollow needle, then you can't understand it.
Another example is Chaperonin GroEL, a human protein. It is shaped like a barrel and holds things inside of it. If you don't know that it is shaped like a barrel, how would you know that it holds things inside it?
Do proteins have only one shape? (Score:2)
My understanding is that the folding processes is rather chaotic. In which case I would expect many different shapes from the same gene, some of which might be useful.
My guess is that the big breakthrough will be to go the other way. Design a shape that does something useful and then be able to create a gene that creates a protein that folds that way.
Re: (Score:2)
My understanding is that the folding processes is rather chaotic.
I don't know if I would call it chaotic. More like wildly expressive and complex.
In which case I would expect many different shapes from the same gene, some of which might be useful.
Most protein sequences do end up in one shape (or in some proteins, there is a section that doesn't really have a shape, it kind of flops around).
The gene is the sequence in the DNA that is the pattern for the protein sequence. Because the pattern can be modified during the construction process, many different proteins can be constructed from a single gene. Figuring out the protein sequence is easy, it's a matter of looking at
Re: (Score:1)
They are competing with IBM's Watson for the most boondoggles.
ERROR (Score:2)
601
Does this mean ? (Score:4, Interesting)
All those years having my screen saver fold proteins was a waste of time?
No it was not a waste fof time (Score:5, Informative)
There are three major protein folding screen saver projects. 1. Rosetta at home. 2. THe World Community Grid human proteome project and 3. folding at home. Additionally there is the game Fold-it.
All of these have had great breaktrhoughs over the years. With the exception of folding at home the rest including the game are based on the early and successful folding algorithm "rosetta". This program is used to design new drugs, new enzymes, and figure out the function of genes by structure. Critical viral proteins and pathway proteins have been identified in those. Folding at home it different, it's more about protein simulation and uses molecular dynamics. It's actually quite bad at predicting protein structure but it models fluctuation dydnamics. This can in favorable case help us understand how protein actually work. For example, how might a covid spike bind and ACH receptor by altering it's shape?
the Alpha FOle version 1 was also derived from Rosetta as well but with AI goodness added. And as you might expect Rosetta itself is now available in an AI flavor. Rosetta has basically come in at the top of every casp competition since 1996 until this last year when it was edged out by the new version of Alpha Fold. The unstated thing about this competition is that it wasn't simply that someone did really well. it's that every algorithm including rosetta improved by the largest leap and bounds ever in the last decade. It just happens that alpahfold did the best this year. The point of the casp competiton isn't to decide "whos best" but rather "what approach works best" then the next time all scientist will incorporate the advance into their algorithms. So AI is a great secret sauce now. and the particular ways AI is applied matter. Unlike the other compeitors Alpha fold did no release their algorithm details (however unlike the other competitors they did just publish a paper on it and are siad to release some of the source code-- it remains to be seen if this is full scientific disclosure as that would require all the training sets and protocols too).
So no you did not waste your effort by donating cycles of cpu time. Tonnes of scientific papers, discoveries and algorithmic improvements came out ot that. People have learned to predict function of proteins from predicted structures rather than measured structures.
The world community grid human proteom project and followions that were even larger produced the same databases that alpha fold just published but 8 years ago. And these became hugely important in understaning the pathways of human biology.
It remains to be seen how much better or complete the new database dump from the alpha fold people was. They have this very annoying tendency not to acknowledge all the owrk that came before them or even work that was fully identical in nature. So if it seems like I'm dissing their work it's because they are not doing all that others have to advance science by sharing their science. It's a net negative when someone says they have a better algorithm but then won't share it with others. That's pighead science. It's like stealing GPL code and not publiching your improvement. I dearly hope they will fully disclose their advances and make me wrong but they didn't do that 3 years ago either. SO we'll see.
On the other hand the work you did with your screen saver was fully shared with the world in data,, results, alpgorithms and protocols. It was piuoneering in that it showed the idea of predicting structures across whole genomes could work, and now we have learned to use that to predict function. SO good for you.