Hi,

I found the following article very interesting, but have no idea how relevant it is to a project such as Rosetta (or indeed if it's news at all!).

Protein 'nanosprings' most resilient found in nature:
http://www.physorg.com/news9900.html

My extremely limited understanding would suggest that if these bonds are significantly stronger than the others in a protein chain, then you could start the simulation with these particular bonds fully folded at the start?

I suspect this isn't news at all, just a research group demonstrating a previously known fact in a new way, possibly in time for the next round of grant allocations (I'm naturally cynical as you may have gathered). Interesting regardless.

-Mike

---

Hi Mike,

Thanks for pointing out this interesting article. Yes, ankyrin repeats form an interesting modular protein domain. They are found in many proteins, and are often involved in protein-protein interactions. What's new in the paper you mention is that people are now harnessing the mechanical properties of ankyrin repeat for nanotechnology, which is really cool.

As for using the structure of ankyrin repeats as a starting point for prediction, it's a bit complicated. This idea would work for any protein that adopts a similar structure as the ankyrin repeats, but for proteins that adopt different folds, it would not work. Even though ankyrin repeats are quite stable, the reason they are stable and fold in their particular pattern, is because they are composed of a certain amino acid sequence that favours that fold. Proteins with different sequences, if you tried to force them into an ankyrin like fold, would not be stable (and Rosetta's energy function would recognize that with poor score).

There are 2 types of structure prediction: homology modelling and de novo (aka ab initio) prediction. For now rosetta@home is performing only ab initio structure predictions (because that is the harder problem). In ab initio prediction, a random starting conformation is used. But the Baker lab also does homology modelling and in this case, you look at the sequence of a protein and if it is similar to something you already know (eg an ankyrin repeat or another type of protein structure), then you do indeed start your simulation from that particular structure.

Cheers,
V.

---

Hi Vanita,

That's very interesting, thanks :-)

---