The rotations are scaled by an annealing factor. As the function
progresses through a give number number of trials, the range of possible
angles decreases proportionally. The starting rotations are completely
random.
Paul.
On 29/03/14 19:35, Oliver Clarke wrote:
Many thanks for the quick response Paul.
So It scales only the translations, not the rotations? Are the test rotations
systematic then, not random? What are the starting values that it scales -
random between 0 and 10 Å in x,y,z for example? (Apologies for all the
questions!)
Cheers,
Oliver.
On Mar 29, 2014, at 3:25 PM, Paul Emsley <[email protected]> wrote:
On 29/03/14 19:19, Oliver Clarke wrote:
Hi all,
I was wondering if anyone could shed some light on what the scale factor in
jiggle fit represents? Empirically, I’ve found that a value of 0.1 with
1000-10000 trials seems to work pretty well for fitting domains in cases where
the initial orientation is quite far from the optimum (where rigid body fitting
fails), but it would help to know what I’m playing around with.
jiggle_scale_factor scales the translations. By default these are 0.1 in each
of the x,y,z directions. In my scripts I use between 1 and 3.
Also, fit_molecule_to_map_by_random_jiggle and
fit_chain_to_map_by_random_jiggle both seem to return 0.0 when called
regardless of whether they find a better orientation - Perhaps it would be
helpful for scripting purposes if they returned the new best score instead? I
am using the python versions of these functions.
Yes it does, yes it would :)
Paul.
