Hi Jose Miguel,
The image is shifted in rounded (integer) pixels to their rlnOriginX and Y
and then translational searches are performed around that origin. This
does not happen with Psi (which would require nasty interpolations, and
psi-searches are almost always exhaustive anyway)
HTH,
Sjors
> Hi Sjors,
>
> A related but a bit different question, If the input star files contains
> the rlnOriginiOffsetX/Y
> they will be used in Relion to 'shift' the raw binary image? What about
> the
> inplane rotation angle?
> It is also taken into account?
>
> Kind regards,
> Jose Miguel.
>
> On Fri, Feb 19, 2016 at 9:04 AM, Sjors Scheres <[log in to unmask]>
> wrote:
>
>> This is almost correct, except for the fact that particles will start
>> being centred at their rlnOriginOffsetX/Y in any 2D or 3D
>> refinement/classification. So if you want a clean restart, you also
>> should
>> modify the origin offsets.
>> HTH,
>> Sjors
>>
>> > Hi Christopher:
>> >
>> > In order to process the unbin data, you just need to modify the
>> > _rlnImageName and _rlnDetectorPixelSize.if you want to recenter the
>> > particles and then re-extract the particles, you have to take care of
>> the
>> > Coordinate X, Y, ,OriginX,Y, and the bin factor.
>> >
>> > Xiaodi
>> > ________________________________________
>> > From: Collaborative Computational Project in Electron cryo-Microscopy
>> > [[log in to unmask]] On Behalf Of Christopher Akey [[log in to unmask]]
>> > Sent: Thursday, February 18, 2016 10:10 AM
>> > To: [log in to unmask]
>> > Subject: [ccpem] scaling problem going from bin2 to full sized
>> particles
>> >
>> > Sjors
>> >
>> > For economy of time I have been doing all the processing of single
>> > particles with bin2 images, and after a 3D classification
>> > run, wanted then to use a particular class and run a 3D refinement
>> with
>> > the bin1 (ie full sized) particles.
>> >
>> > I wrote a script to update the micrograph/particle names, to update
>> x,y,
>> > magnification and 3 defocus parameters in the
>> > existing 3D-class.star file to represent the full sized particles, and
>> > then wrote out a new star with the same header format.
>> >
>> > However, I get a strange scaling error I have not encountered before,
>> but
>> > note that I did normalize the full size particles
>> > when they were extracted from their micrographs.
>> >
>> > At the first Expectation step the job stalled and a file appeared in
>> the
>> > working dir > Mweight.spi
>> >
>> > The cluster slowed down to a load near zero, and the .err output file
>> > contains a very long list of 0 values and a few comments
>> >
>> > at the top:
>> >
>> > exp_thisparticle_sumweight= 0
>> > part_id= 64278
>> > group_id= 8 mymodel.scale_correction[group_id]= 1
>> > exp_ipass= 0
>> > sampling.NrDirections(0, true)= 192 sampling.NrDirections(0, false)=
>> 192
>> > sampling.NrPsiSamplings(0, true)= 24 sampling.NrPsiSamplings(0,
>> false)=
>> > 24
>> > mymodel.sigma2_noise[ipart]= qq
>> >
>> > at the tail:
>> >
>> > mymodel.avg_norm_correction= 1
>> > wsum_model.avg_norm_correction= 0.868591
>> > written out Mweight.spi
>> > exp_thisparticle_sumweight= 0
>> > exp_min_diff2[ipart]= 738.506
>> > ERROR!!! zero sum of weights....
>> > File: src/ml_optimiser.cpp line: 4239
>> >
>> > In thread 0
>> >
>> > Is there another set of params I should have updated or deleted in the
>> > 3D-class.star file?
>> >
>> > thanks,
>> >
>> > C Akey
>>
>>
>> --
>> 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
>>
>
--
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
|