Protein folding solved?

Get your science fix here: research, quackery, activism and all the rest
Beaker
Buzzberry
Posts: 45
Joined: Tue Nov 12, 2019 5:28 pm

Protein folding solved?

Post by Beaker » Mon Nov 30, 2020 10:40 pm

Looks like a spectacular result for AlphaFold, predicting the structure of proteins from their sequence. I am properly amazed by this.

https://www.nature.com/articles/d41586-020-03348-4

User avatar
Grumble
Dorkwood
Posts: 1525
Joined: Mon Nov 11, 2019 1:03 pm

Re: Protein folding solved?

Post by Grumble » Tue Dec 01, 2020 8:22 am

Beaker wrote:
Mon Nov 30, 2020 10:40 pm
Looks like a spectacular result for AlphaFold, predicting the structure of proteins from their sequence. I am properly amazed by this.

https://www.nature.com/articles/d41586-020-03348-4
It’s great to have a good news story. It feels like we’re really starting to make inroads on some hard biological questions, between this and the recent mRNA vaccines.
I know this is vitriol, no solution, spleen venting, but I feel better having screamed, don’t you?

User avatar
dyqik
Stummy Beige
Posts: 3229
Joined: Wed Sep 25, 2019 4:19 pm
Location: Masshole
Contact:

Re: Protein folding solved?

Post by dyqik » Wed Dec 02, 2020 12:12 am

A fairly major issue with this is that it's a closed project that other researchers can't use (and is built on open info/tools).

As such, as far as doing science is concerned, it doesn't exist.

User avatar
Bird on a Fire
Light of Blast
Posts: 4838
Joined: Fri Oct 11, 2019 5:05 pm
Location: nadir of brie

Re: Protein folding solved?

Post by Bird on a Fire » Wed Dec 02, 2020 12:16 am

dyqik wrote:
Wed Dec 02, 2020 12:12 am
A fairly major issue with this is that it's a closed project that other researchers can't use (and is built on open info/tools).

As such, as far as doing science is concerned, it doesn't exist.
Yes, thanks dyqik.

I'm enormously uncomfortable with the privatisation of science that this represents. There are banally un-Philip K. Dick reasons not to want Google to have a monopoly on predicting proteins from DNA.
THINK BIG AND UPEND THE SYSTEM

#ShowYourStripes

User avatar
Bird on a Fire
Light of Blast
Posts: 4838
Joined: Fri Oct 11, 2019 5:05 pm
Location: nadir of brie

Re: Protein folding solved?

Post by Bird on a Fire » Wed Dec 02, 2020 12:19 am

I think a lot of people at Silicon Valley genuinely believe they're doing good work for humanity through this research, and the achievement is undeniably incredible. But I think the well-intentioned researchers are massively naive about the economic system they're benefiting.

I like the idea of Google supporting this kind of research, but it should be free and open-sourced from top to bottom. Like a revolution of the kind of steam-age industrial philanthropy that gets slave-traders streets named after them on modern housing estates.
THINK BIG AND UPEND THE SYSTEM

#ShowYourStripes

User avatar
Bird on a Fire
Light of Blast
Posts: 4838
Joined: Fri Oct 11, 2019 5:05 pm
Location: nadir of brie

Re: Protein folding solved?

Post by Bird on a Fire » Wed Dec 02, 2020 12:22 am

Bird on a Fire wrote:
Wed Dec 02, 2020 12:19 am
it should be free and open-sourced from top to bottom
As should all science, obviously, but this stuff is even less free and more closed-source than most of the rest of what we're doing, and almost certainly more important than much of it. No way should these findings be the private property of the people behind Google+.
THINK BIG AND UPEND THE SYSTEM

#ShowYourStripes

User avatar
Bird on a Fire
Light of Blast
Posts: 4838
Joined: Fri Oct 11, 2019 5:05 pm
Location: nadir of brie

Re: Protein folding solved?

Post by Bird on a Fire » Wed Dec 02, 2020 12:24 am

Bird on a Fire wrote:
Wed Dec 02, 2020 12:19 am
I think a lot of people at Silicon Valley genuinely believe they're doing good work for humanity through this research, and the achievement is undeniably incredible. But I think the well-intentioned researchers are massively naive about the economic system they're benefiting.
I mean, listen to the sack-cloth-and-ashes pieces-to-camera performed by those tech-psychologist guys who came up with the attention-hacking algorithms used to pump unmoderated AI-generated adverts into children's brains.
THINK BIG AND UPEND THE SYSTEM

#ShowYourStripes

User avatar
Bird on a Fire
Light of Blast
Posts: 4838
Joined: Fri Oct 11, 2019 5:05 pm
Location: nadir of brie

Re: Protein folding solved?

Post by Bird on a Fire » Wed Dec 02, 2020 12:38 am

"Google does GATTACA"
THINK BIG AND UPEND THE SYSTEM

#ShowYourStripes

Beaker
Buzzberry
Posts: 45
Joined: Tue Nov 12, 2019 5:28 pm

Re: Protein folding solved?

Post by Beaker » Wed Dec 02, 2020 3:51 pm

It would be fabulous if some university department had the breakthrough and made it open source, but evilcorp got there first.

Suppose Brucker made a better model of diffractometer. Would we expect them to put all that know how in the public domain, or should they sell the box and/or a service of structure determination? Should an inventor of a software solution give their invention away?

I don’t have an answer for that, but I don’t think we can exclude commercial organisations from doing research, even important fundamental research.

Now there is a commercial proof of concept, it ought to stimulate more academic research funding in this field?

User avatar
Gfamily
After Pie
Posts: 2091
Joined: Mon Nov 11, 2019 1:00 pm
Location: NW England

Re: Protein folding solved?

Post by Gfamily » Wed Dec 02, 2020 4:12 pm

I know it's not the same, but ehat happened with the genes that Ventner patented while working alongside the Human Genome Project?
My avatar was a scientific result that was later found to be 'mistaken' - I rarely claim to be 100% correct
ETA 5/8/20: I've been advised that the result was correct, it was the initial interpretation that needed to be withdrawn
Meta? I'd say so!

User avatar
Boustrophedon
After Pie
Posts: 1638
Joined: Mon Nov 11, 2019 3:58 pm
Location: Lincolnshire Wolds

Re: Protein folding solved?

Post by Boustrophedon » Wed Dec 02, 2020 4:17 pm

I know nothing and bearing messrs Dunning and Kruger firmly in mind:

I have never understood why the problem is so very difficult.
The protein is extruded or emerges one amino acid at a time from the ribosome. I cannot imagine that it emerges in a straight line and then folds into the functional conformation all at once, it must therefore start to fold as it emerges and adopts the lowest energy conformation at each step. So the problem breaks down into lots of little local calculations applied in series. Of course there may be rearrangements of two blocks already constructed relative to one another. I know it's a little bit more complicated than that, but earlier attempts at calculating the conformation all at once using supercomputers seem misguided.

What am I missing?
Remember it's only a coup if it's from the coup d'état region of France, otherwise it just sparkling white terrorism.

Beaker
Buzzberry
Posts: 45
Joined: Tue Nov 12, 2019 5:28 pm

Re: Protein folding solved?

Post by Beaker » Wed Dec 02, 2020 4:30 pm

If you completely unravel a protein, it can assemble itself back to an active shape. The reason it’s so hard is generally because there is a mind boggling complexity of possible solutions.

More here:

https://www.eurekalert.org/features/do ... 122204.php

User avatar
jaap
Fuzzable
Posts: 263
Joined: Mon Nov 11, 2019 2:05 pm
Location: Netherlands
Contact:

Re: Protein folding solved?

Post by jaap » Wed Dec 02, 2020 5:03 pm

Boustrophedon wrote:
Wed Dec 02, 2020 4:17 pm
I know nothing and bearing messrs Dunning and Kruger firmly in mind:

I have never understood why the problem is so very difficult.
The protein is extruded or emerges one amino acid at a time from the ribosome. I cannot imagine that it emerges in a straight line and then folds into the functional conformation all at once, it must therefore start to fold as it emerges and adopts the lowest energy conformation at each step. So the problem breaks down into lots of little local calculations applied in series. Of course there may be rearrangements of two blocks already constructed relative to one another. I know it's a little bit more complicated than that, but earlier attempts at calculating the conformation all at once using supercomputers seem misguided.

What am I missing?
That just gives you a local minimum, which is likely not a global minimum. Due to Brownian motion or other interactions, the protein will jiggle about and will likely find even better folding shapes than the one it started off in. With many optimisation problems you can get very good solutions just by doing various small changes. If those small changes are independent things are easy, but if not, i.e. if one small change blocks another, or allows a third, it becomes a labyrinth of small changes that you have to do in the right order to reach the best configuration.

User avatar
Grumble
Dorkwood
Posts: 1525
Joined: Mon Nov 11, 2019 1:03 pm

Re: Protein folding solved?

Post by Grumble » Wed Dec 02, 2020 5:39 pm

jaap wrote:
Wed Dec 02, 2020 5:03 pm
Boustrophedon wrote:
Wed Dec 02, 2020 4:17 pm
I know nothing and bearing messrs Dunning and Kruger firmly in mind:

I have never understood why the problem is so very difficult.
The protein is extruded or emerges one amino acid at a time from the ribosome. I cannot imagine that it emerges in a straight line and then folds into the functional conformation all at once, it must therefore start to fold as it emerges and adopts the lowest energy conformation at each step. So the problem breaks down into lots of little local calculations applied in series. Of course there may be rearrangements of two blocks already constructed relative to one another. I know it's a little bit more complicated than that, but earlier attempts at calculating the conformation all at once using supercomputers seem misguided.

What am I missing?
That just gives you a local minimum, which is likely not a global minimum. Due to Brownian motion or other interactions, the protein will jiggle about and will likely find even better folding shapes than the one it started off in. With many optimisation problems you can get very good solutions just by doing various small changes. If those small changes are independent things are easy, but if not, i.e. if one small change blocks another, or allows a third, it becomes a labyrinth of small changes that you have to do in the right order to reach the best configuration.
Is it analogous to a travelling salesman problem?
I know this is vitriol, no solution, spleen venting, but I feel better having screamed, don’t you?

User avatar
jaap
Fuzzable
Posts: 263
Joined: Mon Nov 11, 2019 2:05 pm
Location: Netherlands
Contact:

Re: Protein folding solved?

Post by jaap » Wed Dec 02, 2020 6:14 pm

Grumble wrote:
Wed Dec 02, 2020 5:39 pm
jaap wrote:
Wed Dec 02, 2020 5:03 pm
That just gives you a local minimum, which is likely not a global minimum. Due to Brownian motion or other interactions, the protein will jiggle about and will likely find even better folding shapes than the one it started off in. With many optimisation problems you can get very good solutions just by doing various small changes. If those small changes are independent things are easy, but if not, i.e. if one small change blocks another, or allows a third, it becomes a labyrinth of small changes that you have to do in the right order to reach the best configuration.
Is it analogous to a travelling salesman problem?
Similar, but much messier.

TSP is a clean mathematical problem. For each instance you can formulate its dual which has the wonderful property that when you've solved both the original and the dual and get the same answer, then you have proved the solutions are optimal. Basically, you bound the optimal tour length from above (by finding solutions to the original instance) and from below (by finding solutions to the dual instance).

Protein folding won't have a dual problem like that, but the basic general properties and reasons for being difficult to solve are the same.

User avatar
bjn
Catbabel
Posts: 938
Joined: Wed Sep 25, 2019 4:58 pm

Re: Protein folding solved?

Post by bjn » Wed Dec 02, 2020 9:27 pm

Bird on a Fire wrote:
Wed Dec 02, 2020 12:19 am
I think a lot of people at Silicon Valley genuinely believe they're doing good work for humanity through this research, and the achievement is undeniably incredible. But I think the well-intentioned researchers are massively naive about the economic system they're benefiting.

I like the idea of Google supporting this kind of research, but it should be free and open-sourced from top to bottom. Like a revolution of the kind of steam-age industrial philanthropy that gets slave-traders streets named after them on modern housing estates.
One big reason this happened is money. I know people at Deep Mind*, they are flush with cash in a way that publicly funded research never is. They can explore a range of problems in a range of ways with a highly motivate and well rewarded team, none of who have to continually interrupt work to chase grants. And they are flush with cash because they can make a mint from it.

*Some deluded recruiter tried to tap me up as Deep Mind CTO a while back, I was tempted for 5 seconds, but it would be well beyond my management skills.

User avatar
bolo
Snowbonk
Posts: 563
Joined: Mon Nov 11, 2019 1:17 pm
Location: Washington DC

Re: Protein folding solved?

Post by bolo » Wed Dec 02, 2020 10:35 pm

If Google weren't a behemoth corporation, it wouldn't have behemoth-level money to throw at this.

If this had to be free and open-source, Google wouldn't expect to make a profit on it and wouldn't want to throw money at it.

You can call the researchers massively naive about the economic implications. You might find that they'd call you massively naive about the alternative.

User avatar
dyqik
Stummy Beige
Posts: 3229
Joined: Wed Sep 25, 2019 4:19 pm
Location: Masshole
Contact:

Re: Protein folding solved?

Post by dyqik » Wed Dec 02, 2020 11:05 pm

Beaker wrote:
Wed Dec 02, 2020 3:51 pm
It would be fabulous if some university department had the breakthrough and made it open source, but evilcorp got there first.

Suppose Brucker made a better model of diffractometer. Would we expect them to put all that know how in the public domain, or should they sell the box and/or a service of structure determination? Should an inventor of a software solution give their invention away?
Possibly. Another model is to contract to build a national research facility for this, with e.g. US Federal cash (via NSF, NIH or similar), which is then operated under something like the "Open Skies" policies that telescopes operate under. Or an academic effort to do the same could be started. Then you are doing science rather than commercial R&D, and enabling the solution of scientific problems.

The "software" solution here includes the massive computing facility required to operate the software, plus probably the assistance and support needed to actually make it solve problems and to interpret the results. But it's a tool that's been demonstrated here, not a scientific breakthrough in itself.

User avatar
shpalman
Stummy Beige
Posts: 3049
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: Protein folding solved?

Post by shpalman » Wed Dec 02, 2020 11:18 pm

It would make a lot of sense to set something up at the EMBL in Grenoble, next to the ESRF where a lot of protein crystallography is done.
molto tricky

User avatar
bolo
Snowbonk
Posts: 563
Joined: Mon Nov 11, 2019 1:17 pm
Location: Washington DC

Re: Protein folding solved?

Post by bolo » Wed Dec 02, 2020 11:26 pm

dyqik wrote:
Wed Dec 02, 2020 11:05 pm
Beaker wrote:
Wed Dec 02, 2020 3:51 pm
It would be fabulous if some university department had the breakthrough and made it open source, but evilcorp got there first.

Suppose Brucker made a better model of diffractometer. Would we expect them to put all that know how in the public domain, or should they sell the box and/or a service of structure determination? Should an inventor of a software solution give their invention away?
Possibly. Another model is to contract to build a national research facility for this, with e.g. US Federal cash (via NSF, NIH or similar), which is then operated under something like the "Open Skies" policies that telescopes operate under. Or an academic effort to do the same could be started.

The "software" solution here includes the massive computing facility required to operate the software, plus probably the assistance and support needed to actually make it solve problems and to interpret the results.
A government-funded facility of this sort would be under enormous pressure to spread its resources around lots of researchers and lots of institutions working on lots of different projects, so it would be unlikely to focus enough on a single problem like protein folding to be competitive with a company like Google, if the company really set its mind to it.

Another problem for AI research in general is that Google et al. are able to offer salaries much higher than universities or government facilities can match. I've been told of major university AI research groups being bought out en bloc. You could argue that universities and government facilities should just be willing to pay more, but that's probably not politically realistic. At least not in the U.S.

User avatar
dyqik
Stummy Beige
Posts: 3229
Joined: Wed Sep 25, 2019 4:19 pm
Location: Masshole
Contact:

Re: Protein folding solved?

Post by dyqik » Wed Dec 02, 2020 11:27 pm

bolo wrote:
Wed Dec 02, 2020 10:35 pm
If Google weren't a behemoth corporation, it wouldn't have behemoth-level money to throw at this.

If this had to be free and open-source, Google wouldn't expect to make a profit on it and wouldn't want to throw money at it.

You can call the researchers massively naive about the economic implications. You might find that they'd call you massively naive about the alternative.
They've thrown a bunch of money at it, but they haven't produced any science with it, or made any scientific breakthroughs. Unless researchers actively working on proteins and genetics that need to get folding results can use it, it's about the same scientific status as that Tesla Roadster heading for Mars, John Glenn's Mercury flight, or a superduper personal telescope that no one can use.

User avatar
dyqik
Stummy Beige
Posts: 3229
Joined: Wed Sep 25, 2019 4:19 pm
Location: Masshole
Contact:

Re: Protein folding solved?

Post by dyqik » Wed Dec 02, 2020 11:32 pm

bolo wrote:
Wed Dec 02, 2020 11:26 pm
dyqik wrote:
Wed Dec 02, 2020 11:05 pm
Beaker wrote:
Wed Dec 02, 2020 3:51 pm
It would be fabulous if some university department had the breakthrough and made it open source, but evilcorp got there first.

Suppose Brucker made a better model of diffractometer. Would we expect them to put all that know how in the public domain, or should they sell the box and/or a service of structure determination? Should an inventor of a software solution give their invention away?
Possibly. Another model is to contract to build a national research facility for this, with e.g. US Federal cash (via NSF, NIH or similar), which is then operated under something like the "Open Skies" policies that telescopes operate under. Or an academic effort to do the same could be started.

The "software" solution here includes the massive computing facility required to operate the software, plus probably the assistance and support needed to actually make it solve problems and to interpret the results.
A government-funded facility of this sort would be under enormous pressure to spread its resources around lots of researchers and lots of institutions working on lots of different projects, so it would be unlikely to focus enough on a single problem like protein folding to be competitive with a company like Google, if the company really set its mind to it.

Another problem for AI research in general is that Google et al. are able to offer salaries much higher than universities or government facilities can match. I've been told of major university AI research groups being bought out en bloc. You could argue that universities and government facilities should just be willing to pay more, but that's probably not politically realistic. At least not in the U.S.
A Government funded facility for protein folding calculations using this technology would do protein folding calculations. In the same way that a government funded facility for radio astronomy does radio astronomy rather than satellite tracking and imaging.

User avatar
Bird on a Fire
Light of Blast
Posts: 4838
Joined: Fri Oct 11, 2019 5:05 pm
Location: nadir of brie

Re: Protein folding solved?

Post by Bird on a Fire » Thu Dec 03, 2020 12:01 am

Yeah I feel like stuff like physics and space get way more huge public projects funded - all those space missions and particle accelerators and supercomputers and things. Biology seems to be expected to survive to a much greater extent as a by-product of commercial research. (I haven't checked the numbers, but I can't think of many similar big public biology projects - even the coronavirus vaccine has been largely funnelled through private corporations, and that was a global emergency)

It's a pretty weird situation when tech companies are spending more money on important breakthroughs in biological research than governments and health-industry spinoffs.

(And yes, I know that some of the reason for government interest in space and physics is to make weapons, but I'm sure they could weaponise biology if they really put their minds to it - or even do good!)
THINK BIG AND UPEND THE SYSTEM

#ShowYourStripes

User avatar
dyqik
Stummy Beige
Posts: 3229
Joined: Wed Sep 25, 2019 4:19 pm
Location: Masshole
Contact:

Re: Protein folding solved?

Post by dyqik » Thu Dec 03, 2020 12:35 am

Bird on a Fire wrote:
Thu Dec 03, 2020 12:01 am
Yeah I feel like stuff like physics and space get way more huge public projects funded - all those space missions and particle accelerators and supercomputers and things. Biology seems to be expected to survive to a much greater extent as a by-product of commercial research. (I haven't checked the numbers, but I can't think of many similar big public biology projects - even the coronavirus vaccine has been largely funnelled through private corporations, and that was a global emergency)

It's a pretty weird situation when tech companies are spending more money on important breakthroughs in biological research than governments and health-industry spinoffs.
Some of the big facilities that look like physics are actually for chemistry and biology - Diamond Light Source, for example.
Bird on a Fire wrote:
Thu Dec 03, 2020 12:01 am
(And yes, I know that some of the reason for government interest in space and physics is to make weapons, but I'm sure they could weaponise biology if they really put their minds to it - or even do good!)
I mean, no one's discussed China engineering a virus in the past year, have they?

User avatar
bolo
Snowbonk
Posts: 563
Joined: Mon Nov 11, 2019 1:17 pm
Location: Washington DC

Re: Protein folding solved?

Post by bolo » Thu Dec 03, 2020 3:29 am

dyqik wrote:
Wed Dec 02, 2020 11:32 pm
A Government funded facility for protein folding calculations using this technology would do protein folding calculations. In the same way that a government funded facility for radio astronomy does radio astronomy rather than satellite tracking and imaging.
A government funded facility for protein folding calculations would look a lot like a government funded facility for any other high-performance scientific computing task, e.g. the National Energy Research Scientific Computing Center at Lawrence Berkeley National Lab, or the Leadership Computing Facilities at Oak Ridge National Lab and Argonne National Lab. All of these have funded at least some work on protein folding. They all regularly fund biology -- COVID-related research, for example, has been a big push for them this year. But they also fund a vast array of other topics, and that broad base is part of why it was attractive for the government to fund them.

Post Reply