Begin forwarded message:
From: Greg Pintilie <gregdp@gmail.com> Subject: Re: [Chimera-users] Fit to segments Date: December 24, 2015 at 9:38:59 AM PST
[…] that’s a good question… some comments below.
Begin forwarded message:
From: Luca Pellegrini <lp212@cam.ac.uk> Subject: [Chimera-users] Fit to segments Date: December 18, 2015 at 4:07:08 AM PST To: chimera-users@cgl.ucsf.edu
Hi, I am using ‘Fit to Segments’ to fit a crystal structure into the appropriate segment of an 18A EM map of a larger protein complex. The structure is one subunit of the complex.
After initial manual alignment of the structure, fitting to the selected segment works well and produces what I believe is the correct orientation of the structure in the map. However, the fitting stats (Corr/At. Incl.) for the wrong hits are almost as good (within a few percent points), even though for some of them the structure is clearly misplaced in the segment.
This can happen quite often especially at lower resolutions, so I’m not too surprised. Try looking at the Z-score as well. For more details see the “Docking Scores” section at http://ncmi.bcm.edu/ncmi/software/segger/docs. It like won’t be too high either in this case though. In such cases it’s often useful to argue that this is the correct fit using other experimental evidence.
Can anyone please comment/advise on this? It looks like fitting and relative statistics are calculated relative to the entire map rather than the selected segment. I am using Chimera 1.10.1, build 40415.
Yes, it is true the scores are calculated using the whole map. One way to use only the selected segment is to extract the densities for that segment first and then do the fitting procedure using that segment only.