Meta questions

Case number:699969-991033
Topic:General
Opened by:laboo
Status:Closed
Type:Question
Opened on:Tuesday, November 8, 2011 - 03:02
Last modified:Wednesday, November 9, 2011 - 20:20

I'm struggling to understand the reasons for fold.it at a very basic level...and I know little about biology/genetics.

1. Why isn't it possible to watch in realtime how nature actually folds a protien? Does it happen too quickly, can we not see things at that level of detail? This seems key to getting people hooked into the site, but I don't see it explained anywhere.

2. Nature must have some sort of algoritm for folding protiens, no? It can't possibly be trying all the combinations that super computers can't get through in a reasonable amount of time. So...why isn't folding testing *algorithms* for folding across lots of different protiens with known solutions, instead of doing one-offs of individual protiens? I mean, isn't the holy grail to figure out Nature's generic folding algorithm? If so, is that just too difficult a task to even try?

Thanks for any help you can offer in this direction. I would love to see this online somewhere.

laboo

P.S. I would have posted this to the forum, but it's telling me I'm not authorized to see that area when I'm logged in. I can see it when I'm not logged in, but can't post.

(Tue, 11/08/2011 - 03:02  |  2 comments)


infjamc's picture
User offline. Last seen 7 hours 15 min ago. Offline
Joined: 02/20/2009
Groups: Contenders

1. Actually, there have been attempts to monitor protein folding in real time. However, since many proteins fold on the timescale of milliseconds, such methods are of limited usefulness.

2. The issue you've mentioned regarding the impracticality of sampling every possible configuration is well known-- it's called "Levinthal's paradox" ( http://en.wikipedia.org/wiki/Levinthal%27s_paradox ). One possible explanation is that protein folding most likely occurs via conserved pathways; in other words, they go through a series of predictable intermediates.

The challenge, however, is translating this finding to computational protein structure prediction and design. After all, representing the biochemistry involved is very computational expensive. When known proteins with similar sequences exist, researchers can get an ahead start by applying homology modeling (i.e. simply copy the known structure for the segments that are highly similar in sequence). But when there are no good templates available, "de novo" techniques (starting from scratch the hard way) must be used. Most likely, this would involve secondary structure predictions and fragment insertions (randomly selecting short segments with similar sequences from a larger database of existing structures to be copied-- which is what the rebuild tool does).

The bottom line is that Foldit is meant use human intuition to further improve the efficiency and accuracy of existing automated methods, which are limited exactly because researchers have trouble figuring out nature's folding algorithm (or at least implementing it).

infjamc's picture
User offline. Last seen 7 hours 15 min ago. Offline
Joined: 02/20/2009
Groups: Contenders

Correction:

The last sentence should read "The bottom line is that Foldit is meant to use human intuition to further improve the efficiency and accuracy of existing automated methods, which are of limited effectiveness exactly because researchers have trouble figuring out nature's folding algorithm (or at least implementing it)."

Sitemap

Developed by: UW Center for Game Science, UW Institute for Protein Design, Northeastern University, Vanderbilt University Meiler Lab, UC Davis
Supported by: DARPA, NSF, NIH, HHMI, Amazon, Microsoft, Adobe, RosettaCommons