Improvements to unwrapping/wrapping for trajectory processing in the BilayerAnalyzer class.

[blakeaw]

The BilayerAnalyzer class assumes that the coordinates of the input trajectory are wrapped by atom and therefore automatically unwraps the coordinates by atom. This can fail when the first frame is not a good reference for unwrapping. This needs to be upgraded to allow for more options of processing the trajectory, such as the opposite scenario where the input trajectory is unwrapped and needs to be wrapped. It would also be good to allow the user to be able to turn off wrapping/unwrapping.
For the currently assumed case (atom-wise wrapped coordinates), it would probably be good to add in a separate input setting for a reference structure; this could default to the first frame when unwrapping used.

A possible long term solution might be to add a transformations protocol (e.g wrapping/unwrapping, structure alignment) that can be set by the user (with proper defaults and conflict resolution based on the analyses being implemented) that is similar to the suggestion in MDAnalysis/mdanalysis#786

[kain88-de]

Hi

A GSoC Student is implementing an on-the-fly transformation protocol in MDAnalysis directly this summer. His own use case are membrane simulations as well. You can find more information on our blog or drop us a line on the mdanalysis devel mailing list.

[blakeaw]

@kain88-de Okay, cool. Thanks for the heads up.

[kain88-de]

The first PR of the project is here

MDAnalysis/mdanalysis#1902