Research team discovers many new twists in protein folding

[JimF]

This was posted on the Rosetta forum, but seems to be at least as relevant here.

Biophysicists have measured protein folding in more detail than ever before, revealing behavior that is surprisingly more complex than previously known. The results suggest that, until now, much about protein behavior has been hidden to science -- happening on faster timescales and with finer changes in structure than conventional methods could detect.

https://www.sciencedaily.com/releases/2 ... 143951.htm

[Nert]

Interesting article.

The findings resolved long-standing discrepancies between past experimental data and molecular simulations, giving confidence to using such simulations to further probe the behavior of membrane proteins.

It sounds like this article confirms the validity of FAH findings versus experimental methods. Am I reading that right ? I'm curious about this discrepancy between simulations (presumably FAH) and experimental data. Can anyone provide a non-scientist explanation of these discrepancies ? Does FAH correctly identify the subtle "intermediate states" described in the article ?

[bruce]

For example, the JILA team identified 14 intermediate states -- seven times as many as previously observed --
...
Better instruments revealed all sorts of hidden dynamics that were obscured over the last 17 years when using conventional technology.

Note the word "observed" I'm going to guess that he's talking about experimentally observed which is not the same as being identified by FAH. The words "using conventional technology" seem to confirm my assumption -- though my assumption still may be wrong.

FAH's general methodology starts a study with projects that start from random configurations rather than a single fully folded or fully unfolded shape. Those random trajectories examine all sorts of different shapes and progress through intermediate shapes OTHER THAN those starting from a single configuration. This random search identifies many intermediate shapes and they extract dwell periods on a similar basis to those he identifies with 8 microseconds. It's computationally "impossible" to find them without this initial phase of random searches.

Once the scientist is convinced of the statistical basis of identify (using his numbers) either 7 or 14 intermediate shapes a secondary study is initiated from each of those intermediate shapes to determine a statistical basis for what I'll call "What happens next if it happens to reach this particular intermediate shape?"

Except for the fact that he seems unaware of FAH's methodology, his conclusions agree with FAH's, such as

If you miss most of the intermediate states, then you don't really understand the system.

Many other in silico (computerized) studies have focused on correlating the initial and final states, ignoring the intermediate details ... and that's an important reason why FAH is so computationally intensive.

The big breakthough described in that article is that in vitro studies are catching up with many of the results that FAH already knows.

[bruce]

Even if you've read the wikipedia article about FAH, look at it again.
https://en.wikipedia.org/wiki/Folding@home

The second diagram on that page shows a number of intermediate states between a denatured shape on the left and a folded shape on the right. The arrows show the probable transitions, with the density of each arror indicating the relative probability of each of those transitions.

(It's actually a lot more complicated that shown. With hundreds of thousands of atoms in a particular project, that diagram would have maybe a million dimensions. The diagram shows only two of those dimensions which makes it more understandable.)