Does new Quantum Folding theory affect @Home efforts?

[markroussey]

Hi folks,

I was wondering if FAH is on top of, or at least takes into account/compensates for the relatively newly published findings from Inner Mongolian U. in China regarding the Quantum states of protein folding, which I can't seem to find any other reference to in this forum. In case this is news, a couple of Mongols have invaded this scene with an explanation for temperature dependence on folding that implies that proteins fold by jumping via quantum transition, rather than mechanically. I can't post links, as this is my first day on the forum, but I think if you google the following terms, you'll see the article in Technology Review that has a link at at the bottom to the originally published paper:

technology review Physicists Discover Quantum Law of Protein Folding

I will post the link upon request, as soon as I am enabled to.

What do you think? Does this shift the paradigm? Does this have implications for the current software, i.e. does it need to be re-written from scratch? I'd like to know before I fire up the CUDA cores on my Nvidia card.

Thanks in advance for any qualified response!

- Mark

[justjohn]

hi Mark,
I see you haven't received a response here, so I'll comment. But I'm "un-qualified", with just an old BS in CompSci.

First, the links to the paper and blog posting:
http://arxiv.org/abs/1102.3748
http://www.technologyreview.com/blog/arxiv/26421/

I'm not familiar with the arxiv site, but it seems to be a spot for sharing papers, without peer review?
http://arxiv.org/help/primer

I've done computer support for some Political Science journals (peer reviewed), the way the process usually works is that a paper comes in, the editor makes sure it is "reasonable" (on topic and not looney) and then is sent out to ~three reviewers. I think two thirds need to give a thumbs-up before the paper is published. And they might bump it back to the author for clarification.

Two complaints about this peer review process are that it slows things down, and adds expense (thus limiting distribution to paying subscribers). I think Arxiv is a fast & free solution, so that is great. My Big10 uni seems to have made a donation to them, so they must have some credibility.

Since this paper was just published about two weeks ago, I doubt that the F@H project is going to make huge changes right away. I've only had a few semesters of CEM, and was a poor student at that. But to me, this paper seems sort of off-the-wall (if I'm understanding it): a large protein molecule just sort of pops from one shape, to another? What about all the energy inputs needed to change state? How does the protein know the final shape it is supposed to move into?

Personally I look at the published research Pande Group has done (78 papers), the quality of the F@H software I'm running on dozens of computers (damn fine) and the future path for folding science. I'm going full speed ahead for F@H.

You should make up your own mind, but I'd suggest getting your CUDA cores hot.

[7im]

Pande Group's simulations have won awards for their theoretical prediction of proteins shapes matching actual folded proteins shapes.

So I'd guess that either PG's methods already account for this new "quantum" discovery, or the the discovery is not what it seems...

I agree. Fire up the folding hardware.

[rbpeake]

I believe it was Carl Sagan who said that "extraordinary claims require extraordinary proof", so I would agree with the above comments that this claim may not yet have been fully vetted by the scientific community.

[VijayPande]

The temperature dependence of folding is well-explained by the entropic nature of hydrophobicity and the nature of the entropic bottleneck involved in a protein finding its way to the folded state. The authors seem unaware of this basic, well-established fact and have what looks to be a pretty strange explanation for it. Unfortunately, this paper is not peer reviewed at all and I think this sort of simple fact would naturally have come out in that process.

[markroussey]

Thanks to all for taking their time to reply, and especially for Dr. Pande to take his personal time to respond. Justjohn - thanks for the links on the arvix article; here's the link to the actual paper:

http://arxiv.org/ftp/arxiv/papers/1102/1102.3748.pdf

My interest in tracking down the viability of their claim was piqued (in a paper that is actually pretty short - only 16 pgs, of which the first six make their case) when their model seemingly correctly forecast folding times for a variety of proteins. While these guys do not outright dispute the possibility of the classical physics explanations for the highly variable folding rates, they do seem to have a direct formula to explain them, whereas we are still deriving classic physics formulas to explain the variety of folding states, no? And, they pose an interesting question - one that does challenge the status quo, and begs for a response - from a qualified peer, such as you:

"...It is well known that the fluorescence and phosphorescence are phenomena closely related to protein folding. Since no one doubts that fluorescence and phosphorescence could only be understood in terms of the quantum transition between molecules, why should the protein folding study be divorced from the framework of quantum theory?"

Is that a valid statement - "...fluorescence and phosphorescence could only be understood in terms of the quantum transition between molecules..." - ? If so - not that you haven't more than a few other items on your to-do list, but - how about looking at their paper? I'm just not qualified to comment on it. And, who knows - maybe they're onto something. What do you think?

- Mark Roussey

[ihaque]

My interest in tracking down the viability of their claim was piqued (in a paper that is actually pretty short - only 16 pgs, of which the first six make their case) when their model seemingly correctly forecast folding times for a variety of proteins.

I don't see anything in this manuscript that "forecasts" folding times. The only data presented are in Table I: columns 1, 2, and 3 are linear fit parameters to experimental data (standard methods) and column 4 is the measurement temperature. Column 9 is a parameter relating to the size of the protein chain (I think). Columns 5-8 are not predictions: they are parameters of their model which they fit to the experimental data. That in itself isn't awful (you usually have to fit parameters somewhere); what's bad is that ALL their results are from model fits. Correct statistical practice dictates that you train your model (fit parameters) on a subset of the data, and then evaluate your model on data on which your model was not trained. Otherwise, all you're doing is testing your ability to read from a table, not predict anything. It's possible to invent essentially any sort of wacky model and get it to fit data perfectly, if given enough free parameters and no evaluation set.

While these guys do not outright dispute the possibility of the classical physics explanations for the highly variable folding rates, they do seem to have a direct formula to explain them

I see no such formula (or at least, attempt to apply it). If you look at columns 5-8, the numbers are all over the place, and they have no explanation for how you might derive this without actually doing the experiment (at which point, you're not really generalizing or predicting anything).

"...It is well known that the fluorescence and phosphorescence are phenomena closely related to protein folding. Since no one doubts that fluorescence and phosphorescence could only be understood in terms of the quantum transition between molecules, why should the protein folding study be divorced from the framework of quantum theory?"

To paraphrase Wolfgang Pauli (as long as we're talking about quantum :p), this isn't right; it's not even wrong. Fluorescence and phosphorescence are connected to protein folding only insofar as they are often used as indicators of the folded state, because the fluorescence from a given atom or set of atoms is highly sensitive to its local environment (eg, whether there's protein or water nearby). They don't directly cause protein folding. It's true that fluorescence is inherently a quantum effect (electrons get excited to higher energy states and emit photons when they drop to the ground state), but that doesn't imply that quantum effects are actually relevant to folding transitions. Let's put it another way: fluorescence is quantum, so the operation of that fluorescent light tube above you can only be understood in terms of quantum transitions between electronic energy levels. Why should the study of installation of fluorescent light tubes in buildings be divorced from the framework of quantum theory?

Note that I'm not claiming that quantum is irrelevant (though I don't think it's been shown to be important to folding) - just that nothing this manuscript said actually justifies that it is.

[markroussey]

Thank you, ihaque, for a succinct and complete reply. Your points are well put. Also, know that another esteemed individual in your field, a scientist from U of I - Urbana Champaign, who has received numberous awards and honors for computational biology, reflected similar thoughts. (I haven't received permission to name him yet). His take on the paper:

"...in a nutshell, their theory has no quantum effects in it; they quickly take expectation values of operators, then do classical statistical mechanical sums on them."

So - thank you all for putting this apparent non-issue to bed. It's hard to seperate the b.s. from the real thing at my level - technical enough to get completely lead astray. I will proceed to gear up for Folding @ Home. I am sorry if I raised any hackles in the process of vetting this claim.

- Mark

[ihaque]

It's hard to seperate the b.s. from the real thing at my level - technical enough to get completely lead astray.

A good first step when you're not familiar with the field is to immediately ignore anything in arXiv. I personally think preprint archives are a great service for the scientific community, but it's important to realize that the papers are NOT peer-reviewed; more or less anyone can deposit a manuscript of complete bollocks into arXiv if they so desire. Peer-review isn't perfect, but it is a mark that some people who are familiar with the field have taken at least a brief look at the paper.

I think science reporting does the lay public a disservice when it reports on arXiv papers uncritically, for precisely the reasons we've seen in this thread.