Hi Austin,
The relion_project programme can be used to make 2D projections of a 3D
map. The (new) --nr_uniform command controls how many you make. Make sure
to not use the full resolution in the autopicking, as you may become
susceptible to 'Einstein-from-noise' artefacts. We typically use 20A
resolution for this, you can control it in the autopicking job.
Overall, you might get a few more particles, but I don't expect great
improvements, provided you have already used 2D class averages in the
autopicking. But one can always try. :-) With the GPU-acceleration,
autopicking with many references should still be quick.
HTH,
Sjors
> I have a pretty good 3D structure of a protein we are working on with a
> lot of heterogeneity in a certain section. I have used multiple
> refinements, masking, and refinements with finer angular sampling to get a
> structure that has most of the features we are looking for, but it
> represents a very small percentage of the particles originally picked, and
> is at >10 A resolution. I'd like to increase that number of
> particles/resolution by using this 3D consensus to make 2D references and
> then run that through autopicking. Hopefully that will net me more high
> quality particles that contain the subunits of interest.
>
> Is there a way to do this Relion? Chimera?
>
> Also, is this introducing bias in any way? The 2D classes would be
> generated from a 3D structure that was generated from the same raw data,
> so I wouldn't think so, but any advice on that would be great.
>
> Thanks!
> Austin Dixon
> Rockefeller University
>
--
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
|