Hi Nathan,
Jelly body refinement is basically a way of stabilising refinement, helping to achieve a more robust estimate of the curvature of the likelihood function, thus leading to a more robust search direction. This is similar to other regularisers used in refinement, e.g. geometry restraints, B-value restraints, external restraints. The idea is that such restraints should make refinement should be more robust to effects such as noisy data.
There are two fundamental differences with jelly body restraints: (1) these "restraints" do not affect the likelihood function, or the gradient, but are only applied to the 2nd derivative, thus only affecting the curvature of space; (2) rather than being based on externally-acquired prior knowledge, the prior knowledge is the current state, i.e. the current interatomic distances between close atoms.
The sigma controls the effective weight of the jelly body "restraints" relative to the other terms, geometry, x-ray, etc. Low sigma = strong weight. Try sigma = 0.01 - that is the current favourite value. But play around with the sigma and use what works best! Hope that helps.
Regards
Rob
On 23 Aug 2012, at 18:27, Nathan Pollock wrote:
> Dear experts,
>
> Could someone explain what it is exactly that jelly body refinement
> does? I think that I understand it intuitively but want to make sure.
> In the same vein, what does jelly body refinement sigma parameter
> control? I.e., in comparison to the default sigma = 0.02, does sigma =
> 0.1 make body more or less like a jelly fish?
>
> Thanks!
>
> - Nate
|