Hi Christos,
Some thoughts about differences: the reconstruction algorithm including
the CTF-correction in SPARX might be different; SPARX may not do the
postprocessing in the same way (i.e. with phase-randomised correction
for effects of masking on the FSC), and SPARX does not do
probability-weighted angular assignments as in RELION.
To answer your follow-up questions:
1) yes, you can transfer to other packages, but often results do get worse.
2) inside the refinement RELION uses probability-weighhted (smeared if
you like) angular assignments. The output angles and translations are
only the optimal ones. And no: you cannot output the smeared
assignments, as those are very long vectors for each particle they are
only kept in RAM.
3) I would expect the biggest effects on just resolvability of the map,
not so much actually different structures.
HTH,
S
On 03/25/2014 09:07 AM, Dr. Christos Gatsogiannis wrote:
> Dear Sjors,
>
> i'm a bit confused concerning exchanging parameters between RELION and
> other programs. i used successfully RELION to obtain an atomic
> resolution structure and i wanted to test how this dataset behaves
> also with other programs, we use in the lab. I wrote a python script
> for example to convert the RELION alignment parameters (from the MRC
> convention) to SPARX parameters, but then when i calculated a
> reconstruction with SPARX the resolution was 0.5 A worse.
> Can you please give me any hints where might this be coming from?
> After all, it's the same dataset, with the same parameters (alignment
> and CTF) and the FSC was calculated between an odd and even map, using
> the mask, that was also used in RELION.
> Just to be sure, the alignment parameter RELION outputs, are also the
> parameters that are used in the reconstruction within RELION, so it
> would be possible to transfer them to other programs,
> or RELION alters them before reconstruction? (maybe due to gaussian
> smearing)? The problem that i have is that the SPARX map is not only
> less well resolved, but also shows some differences in comparison to
> the RELION map, which does not really make a lot of sense me..Or do
> you think that weighting of the particles that is performed in RELION
> but on in SPARX, might have such a great impact, on the final outcome?
>
> I m looking forward for your reply and any comments/advices,
>
> Best Regards,
>
> Christos
>
> -------------------------------------------------
> Dr. Christos Gatsogiannis
> Max Planck Institute Molecular Physiology
> Raunser Lab
> Otto Hahn Str. 11
> 44227 Dortmund
>
> Tel.: (+49)(231)133-2722
> http://www.mpi-dortmund.mpg.de
> --------------------------------------------------
>
>
--
Sjors Scheres
MRC Laboratory of Molecular Biology
Francis Crick Avenue, Cambridge Biomedical Campus
Cambridge CB2 0QH, U.K.
tel: +44 (0)1223 267061
http://www2.mrc-lmb.cam.ac.uk/groups/scheres
|