3D Refinement extrange behaviour

Submitted by rogelio on Thu, 2014-10-23 21:28

Forums

Frealign

Hi there,

I am using the latest version of Frealign 9.09 (I also got this with 9.08) and I am having a extrange behavior during my 3D refinement. I used data binned by 2 (pixel size 2.0A/px) and obtained a structure that I would like to keep refining with my unbinned data. I copied the last .par file from my refinement and I updated the magnification column and used that as the initial parameters for new cycles of refinement with my unbinned data.
It turns out that after particle 1500 the scores values dropped dramatically resulting in reconstructions with lower resolution than the ones obtained with my data binned by 2. Any idea of what could be happening? I have attached a portion of my par file

Thanks so much

Rogelio

Axel Thu, 2014-10-23 22:32

Hi Rogelio,

The new frealign should not require any modification of the par file when you go from binned to unbinned data. Just change the Dstep parameter to one half of what it was with binned data, and resume refinement.

Axel

niko Fri, 2014-10-24 01:16

You also need to change the pixel size to half the previous value when changing the binning. There should be no change to the .par file. Remember that the magnification should be the value at the detector, not the value displayed by the microscope. It usually has to be calibrated to be accurate. It remains the same no matter what binning you use.

rogelio Sat, 2014-10-25 18:31

Thanks a lot both for your answers, I am in the process of refining the unbinned data. That should solve my problem.
Just for completeness .. would it be different to modify the Dstep and pixel size in the mparameters file vs modifying by 2 the magnification in the parfile and the pixel size in the mparameters file?

I understand it is easier to modify the mparameters file, just want to understand a bit more about how frealign is working,

thanks again,

Rogelio