StalinTheCat

Member
Oct 30, 2017
720
From Nature:

DeepMind's program, called AlphaFold, outperformed around 100 other teams in a biennial protein-structure prediction challenge called CASP, short for Critical Assessment of Structure Prediction. The results were announced on 30 November, at the start of the conference — held virtually this year — that takes stock of the exercise.

"This is a big deal," says John Moult, a computational biologist at the University of Maryland in College Park, who co-founded CASP in 1994 to improve computational methods for accurately predicting protein structures. "In some sense the problem is solved."

"It's a game changer," says Andrei Lupas, an evolutionary biologist at the Max Planck Institute for Developmental Biology in Tübingen, Germany, who assessed the performance of different teams in CASP. AlphaFold has already helped him find the structure of a protein that has vexed his lab for a decade, and he expects it will alter how he works and the questions he tackles. "This will change medicine. It will change research. It will change bioengineering. It will change everything," Lupas adds.

In some cases, AlphaFold's structures predictions were indistinguishable from those determined using 'gold standard' experimental methods such as X-ray crystallography and, in recent years, cryo-electron microscopy (cryo-EM). AlphaFold may not obviate the need for these laborious and expensive methods — yet — say scientists, but the AI will make it possible to study living things in new ways.

Proteins are the building blocks of life, responsible for most of what happens inside cells. How a protein works and what it does is determined by its 3D shape — 'structure is function' is an axiom of molecular biology. Proteins tend to adopt their shape without help, guided only by the laws of physics.

A lot more at the source directly: https://www.nature.com/articles/d41586-020-03348-4

This is groundbreaking and an amazing achievement.
 

Solace

Dog's Best Friend
Banned
Oct 29, 2017
1,920
I have no idea what any of this means but YAY! for scientific advancement!
 

SwampBastard

The Fallen
Nov 1, 2017
11,156
"I think it's fair to say this will be very disruptive to the protein structure prediction field. I suspect many will leave the field as the core problem has arguably been solved," he says. "It's a breakthrough of the first order, certainly one of the most significant scientific results of my lifetime."

Amazing stuff. Good job, Google.
 

Voltaire

Member
Sep 13, 2018
391
Well that was one my potential PhD thesis ... I guess that's that. If the problem is solved as described in the article that's awesome news.
 

Tuorom

Member
Oct 30, 2017
10,998
That's pretty crazy to hear someone say "well, the problem is solved" for something like this.

It feels momentous.

The ability to accurately predict protein structures from their amino-acid sequence would be a huge boon to life sciences and medicine. It would vastly accelerate efforts to understand the building blocks of cells and enable quicker and more advanced drug discovery.
 

SRG01

Member
Oct 25, 2017
7,031
No, they published in Nature

(I'm sorry for this terrible joke)

LOL

This is great, my brother is also getting into this area and the potential has been nothing short of amazing. This should greatly accelerate development of proteins and such, as they can simply dump it into the AI before ordering the samples.
 

GYODX

Member
Oct 27, 2017
7,276
It's hard to overstate what a monumental achievement this is.

Deep learning is awesome.
 

captmcblack

Member
Oct 25, 2017
5,115
I hope that the benefits to human life/longevity that can come from these studies can happen at least before my lifetime's over (so somewhere within the next 30 to 60 years, maybe).
 

LinkStrikesBack

One Winged Slayer
Member
Oct 27, 2017
16,508
I fully endorse computationally minded people replacing biologists.

The result is nice too I guess.
 
Oct 27, 2017
797
Berkeley, CA
Isn't protein folding an NP-hard problem? If the problem is considered "solved", what does that imply about NP problems? Can we use this method to solve other NP problems since they reduce to each other?
 

KDR_11k

Banned
Nov 10, 2017
5,235
There's a manual protein folding game that found that humans are naturally adept at this kind of stuff and can solve it faster than folding@home can but of course that's not automatic.
 

TheRagnCajun

Member
Oct 29, 2017
590
Honest question here: is simulating protein folding actually getting us anywhere? I keep hearing about advances in computation but I have never heard of the results actually helping cure/prevent one disease.
 

GYODX

Member
Oct 27, 2017
7,276
Isn't protein folding an NP-hard problem? If the problem is considered "solved", what does that imply about NP problems? Can we use this method to solve other NP problems since they reduce to each other?
They're only using the term "solved" in the informal sense, not in the computational complexity sense.
 

Damaniel

The Fallen
Oct 27, 2017
6,548
Portland, OR
Isn't protein folding an NP-hard problem? If the problem is considered "solved", what does that imply about NP problems? Can we use this method to solve other NP problems since they reduce to each other?

Even if it is (and I haven't gone and looked, though it seems reasonable), there are plenty of cases of NP-hard problems where reasonable (though not always optimal) solutions can be found for many - but not necessarily all - instances of the problem.