 |
 |
Dear Xavier, dear all,
good points indeed. I guess the aim is to refine an atomic model against a cryo-EM map so that the atomic model describes the experimental data well, aiming for a good final geometry, refined B-factors etc. of the atomic model.
Here is a description of what we recently did for one of our structures:
https://www.nature.com/protocolexchange/protocols/6365
HTH,
Bruno
From: CCP4 bulletin board [mailto:[log in to unmask]]
On Behalf Of Israel Sanchez
Sent: Friday, August 31, 2018 2:32 PM
To: [log in to unmask]
Subject: Re: [ccp4bb] cryo-EM
Dear Xavier,
On top of Rob's excellent suggestion, I would recoment you to have a look at this paper by Alan Brown and co-workers (including Rob himself):
We normally performs an initial real-space refinement with phenix.real-space followed by two step of reciprocal space refinement with Refmac. In the first Refmac run we normally use secondary structure restrains generated by ProSmart for
proteins and LibG for nuclei acids. In the second Refmac run, we try to release the restrains power and also play around with the matrix governing the weight between the experimental data and ideal symmetry. You can monitor when are you entering over-fitting
conditions employing the half maps normally output by Relion (or the program you use for cryoEM data processing). More specific examples here:
and here:
Finally, I would not recommend at all playing around with the phases of a cryoEM map. If you have a region which is less visible due to poorer local resolution, you may blur the map computing a post-precessed map with a lower Bfactor that
the one automatically assigned by Relion or low pass filter the same map a bit. Or even combine both approaches. Not sure what the big guys in the field think about tinkering with the phases of a cryoEM map but I feel this is still highly controversial. May
be Garib or Pavel think differently?
Hope this help.
El vie., 31 ago. 2018 a las 8:02, Rob Nicholls (<[log in to unmask]>) escribió: