[phenixbb] Detrmining pixel size

Reza Khayat rkhayat at ccny.cuny.edu
Sun Jul 8 13:37:58 PDT 2018


A 1.38Å vs 1.40Å pixel size will make a tremendous difference for complexes like viruses.  Imagine an image reconstruction of a virus that is 200 pixels in diameter -pixels are an EM's readout, not distances.  At 1.38Å/pixel this is a diameter of 276Å, but at 1.40Å/pixel this is a diameter of 280Å.  A 2Å expansion is certainly biologically important.  So an incorrect pixel size of 0.02Å will give the perception that a biologically important phenomenon has occurred.  Thus, despite acceptable refinement statistics one can make incorrect conclusions based on incorrect pixel size.  I'm dealing with this problem when comparing my image reconstructions from different scope to crystal structures. 

This problem is amplified as the virus size increases. 

Best wishes,

Reza Khayat, PhD
Assistant Professor
City College of New York
Department of Chemistry
New York, NY 10031

From: phenixbb-bounces at phenix-online.org <phenixbb-bounces at phenix-online.org> on behalf of Pavel Afonine <pafonine at lbl.gov>
Sent: Sunday, July 8, 2018 3:59 PM
To: anaa2; PHENIX user mailing list
Subject: Re: [phenixbb] Detrmining pixel size

Hi Noor,

> Does anyone know how to accurately determine the voxel size of an EM
> map? If you are slightly off (i.e. 1.38 compared to 1.40 pixel) does
> this affect real space refinement in PHENIX?

if it's not too much off then I would not expect much difference in
results except map-model correlation (CC) may be worse.. This is because
phenix.real_space_refine works such that no matter what fit to the map
is the model geometry must stay within acceptable limits.


phenixbb mailing list
phenixbb at phenix-online.org
Unsubscribe: phenixbb-leave at phenix-online.org

More information about the phenixbb mailing list