Dear Tim, On Mon, 2011-04-11 at 10:44 +0200, Tim Gruene wrote: > since you pointed it out I wonder if there is any reasonable (i.e. > w.r.t. data > error/ resolution) difference between the interpolated values and the > calculated > value. I actually doubt that
That should depend on the quality of interpolation primarily (i.e. grid used, resolution, etc). With a 1.9A test case the discrepancy runs somewhere around 3%, which likely is within experimental error. This is using sftools to calculate the map and mapman peek command, all with defaults. Direct calculation versus interpolation looks like this http://tinyurl.com/67f79oe For the record, getting this via mapman is not entirely trivial. Atoms should be inside the map (expanding map to the full unit cell and shifting all the atoms by symmetry inside the (0,1) range was my solution). Also, it fails for some atoms (~0.5% in this case) with "Spline interpolation error", in which case the output pdb file has ****** instead of B-factor (perhaps switching to interpolation can fix that). Also for the record, I believe there is a bug in sftools that messes up the expansion to P1, but only for some space groups/circumstances. For example, another test case in P3121 looks like this (~30% average discrepancy) http://tinyurl.com/69wqhvp but upon closer inspection, the maps after expansion into P1 (or those calculated with sftools FFT command) show a lot of noise. Cheers, Ed -- "Hurry up before we all come back to our senses!" Julian, King of Lemurs
