[Chimera-users] Ways to minimize data loss during resampling?
goddard at sonic.net
Mon Apr 13 13:47:37 PDT 2015
One approach is to align the maps and save a Chimera session which keeps that alignment, no resampling. That doesn’t let you open the aligned maps in any other software than Chimera so it is a limited solution.
Usually maps are saved at a grid spacing of about 1/3 the resolution, for example 1 Angstrom grid spacing for a 3 Angstrom resolution map. With that amount of oversampling, interpolation on a new grid I think usually does not cause much loss of detail. I don’t think any other interpolation schemes that end up with a map on a different grid are likely to produce a significantly better result than the tri-linear interpolation done by Chimera. The basic problem is if you have large density swings from one grid point to the next then you will lose detail, and if you have slow variation from one grid point to the next then any reasonable interpolation will be ok.
> On Apr 8, 2015, at 7:15 AM, Oliver Clarke wrote:
> When one has multiple maps corresponding to the same protein in different conformations, it is convenient to save aligned copies of each map, for easy comparison later or for transfer to other programs.
> This requires interpolating the map to be aligned on the grid of the map to which one is aligning, e.g. using “vop resample x ongrid y” in Chimera. This step involves a certain degree of data loss, which is mostly insignificant but occasionally affects the quality of the resultant map in a way that is visually noticeable.
> Does anyone have any tips for minimizing this data loss, or is that impossible? E.g., if I oversample the map to which I am aligning by lets say 4 times, resample my map on that grid, and then save at gridstep 3 will that help, or is the data loss an unalterable consequence of the interpolation procedure?
> Chimera-users mailing list
> Chimera-users at cgl.ucsf.edu
More information about the Chimera-users