Is there some level of "cheating" or perhaps prior computing that factors into the docking WUs? The reason I ask is that it bothers me that the orientation of the two proteins always seems to be pretty close to the native just after initialization. Is it just that the orientations are decided during those first 500 steps that go by so fast and I don't see it happen or is the orientation already somewhat predetermined?

---

Hi, SOAN. You are right! The docking WUs we are currently running on R@H are sampling round native comlex orientation. This is "cheating" in the sense that we have already known the answer(native complex). But it is an important test for our sampling strategy and energy function because if we can not find the correct conformation in a local search, there is no hope to succeed in a global search. This type of test is called "small perturbation" by us and usually the searching radius is within 30A to the native complex (after translation and rotation). In docking, we also have a blind test competition called CAPRI, similar to CASP for structure prediction. In that competition, we do not know the native structure beforehand and we will have carry out a global docking search to cover the whole space. After making docking compatible with Rosetta@Home, we are looking forward to running jobs on R@H for the next round of CAPRI.

For the actual docking protocol, the first 500 steps do go very fast because the search is done in a low-resolution stage where we just want to find a rough shape fit and after then the long refinement in full-atom stage starts from this orientation. The rigid-body movement in the refinement stage is very small as Rosetta is trying very hard to optimize detailed interactions across the docking interface.

I am happy to answer more questions...

Is there some level of "cheating" or perhaps prior computing that factors into the docking WUs? The reason I ask is that it bothers me that the orientation of the two proteins always seems to be pretty close to the native just after initialization. Is it just that the orientations are decided during those first 500 steps that go by so fast and I don't see it happen or is the orientation already somewhat predetermined?

---

For us laypersons... is a "global search" where you have to test for fit all the way around both proteins? ...and a "local search" where you already know, or are assuming the docking will center near a given point? Sort of like knowing which continent you will land on, and then continue the lowest elevation search from there?

---

Add this signature to your EMail:
Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might!
https://boinc.bakerlab.org/rosetta/

Exactly right. In real life, we are trying to collect any information (mostly from other experiments) which may help to locate a likely starting point and from there we can carry out such "local search". However if such information is not avaiable at all, we will have to try all the possible ways by which two proteins fit together and that is the "global search".

For us laypersons... is a "global search" where you have to test for fit all the way around both proteins? ...and a "local search" where you already know, or are assuming the docking will center near a given point? Sort of like knowing which continent you will land on, and then continue the lowest elevation search from there?

---

Very interesting. Thank you for the explanations, Chu.

---