PDB 1ten protein folding

[MrHarrytubing]

this is like porn for biological scientists :D

[empyrean101]

How do you render images from an atomic coordinate file? Raster3D, OpenGL?

[benlablondon]

I've written an opengl protein renderer, which is useful for watching the simulation as it proceeds.

[dhfrman]

r u able to use bluegene IBM'S protein folding program

[benlablondon]

I don't think IBM have their own protein folding software, they are working with academics to simulat proteins using molecular dynamics. This involves simulating all atoms, and interactions between them, very very slowly, to get as accurate a model as possible.

[benlablondon]

This may be the ultimate solution to the protein folding problem, however it is currently very slow (months!), can only be used to fold quite small proteins, and even then is sometimes quite inaccurate. The world's most powerful computers are used to perform molecular dynamics, as well as using distributed processing such as folding@home.

[benlablondon]

We aim to simulate the folding of smallish proteins (80-150 amino acids) in around an hour using a single desktop PC. Note that this is not to disrespect or underestimate the work of IBM, which I'm sure will be quite amazing!

[TexasRounder]

I am very interested to learn more about what mathematical techniques you may know about that are being used to analyze this problem?

[benlablondon]

There is a good overview on Wikipedia (search for Protein structure prediction) of general methods for protein structure prediction and modelling. We're aiming to publish a paper on our method fairly soon, and I will post a link to it when that is out!

[subwaysnacker1]

Can you comment on what software was used to simulate the folding of the protein and especially what software was used to compile the folding data into this video. This looks amazing!

[benlablondon]

The software to simulate folding, and the software for making the video, were both developed by myself: together they are called poing. I hope to release this software in the near future under an open source license - pending a publication.

[raznshizzle]

This is cool. I don't get why folding from a complete unfolded state (as shown) should follow the same pathways, or have the same final structure as folding while being synthesised though.

[benlablondon]

Regarding the final structure, the theory goes that a protein will find its native, most thermodynamically stable state from any other state. So it should be irrelevant whether it starts in a complete denatured state, or if it is folding whilst it is being synthesized.

[benlablondon]

Regarding the folding pathway, you're right that it will be different for a denatured protein versus a protein being synthesized. Exactly how different we don't know, because the action of chaperones can delay certain aspects of folding until synthesis is complete. This is the subject of some of my research.

[johnesco]

The video is about 3 minutes long. How long does this take is real time? I'd imagine a split second.

[benlablondon]

The real time for protein folding is a tiny fraction of a second. A rule of thumb is about 10ns per residue, although beta-strand based proteins will be slower than this. This protein is 90 residues, so that's at most 900ns, or about a millionth of a second.

[benlablondon]

In addition, proteins generally fold while they are being synthesized, and synthesis is very much slower than folding, so the actual overhead of folding is minimal compared to synthesis. The processing time that went into the 3 minutes of simulation you see here is about 10 minutes (not including rendering!).

[yanman1586]

Cool.