Small group sizes may lead to unstable scale estimation, which may
indeed explain your trouble. The refine program will have warned you of
these dangers in the stdout.
In the upcoming 1.3-release (hopefully available for beta testing in a
few months), the procedure to group micrographs will be much easier
(i.e. just providing the number of groups and then clicking a button).
best,
Sjors
On 01/28/2014 09:37 AM, Ludovic Renault wrote:
> ok, thanks for your reply.
> My case is a bit different then.
> Creating a grouped star file by defoci has removed the errors.
> I'll see if the classification actually finishes now.
> Cheers.
>
>
> On Mon, Jan 27, 2014 at 10:51 PM, Amedee des Georges
> <[log in to unmask] <mailto:[log in to unmask]>> wrote:
>
> My reference was not the right size compared to my particle
> images. It would run fine for two iterations but then the
> reference volume would start to diverge, probably giving little
> signal to align to, hence the error during the expectation step.
> It took me a while to realize that my reference volume was off
> because of a program bug.
>
> --
> Amedee des Georges
> Frank lab
> Howard Hughes Medical Institute
> Dept. of Biochemistry and Molecular Biophysics
> 650 W 168th St, P&S Black Building room 221
> New York, NY 10032
> (T) 212 305-9527
> (F) 212 305-9500
> (E) [log in to unmask] <mailto:[log in to unmask]>
> http://franklab.cpmc.columbia.edu <http://franklab.cpmc.columbia.edu/>
>
>
>
> On Jan 27, 2014, at 5:49 PM, Amedee des Georges
> <[log in to unmask] <mailto:[log in to unmask]>> wrote:
>
>> My reference was not the right size compared to my particle
>> images. It would run fine for two iterations but then the
>> reference volume would start to diverge, probably giving little
>> signal to align to, hence the error during the expectation step.
>> It took me a while to realize that my reference volume was off
>> because of a program bug.
>>
>> On Jan 27, 2014, at 11:33 AM, Ludovic Renault
>> <[log in to unmask] <mailto:[log in to unmask]>> wrote:
>>
>>> Hi Amedee,
>>> I'm having the exact errors you described.
>>> I also change the value to 1 and now am having small or negative
>>> scale error at 3rd iteration of 2d classification.
>>> Did you manage to get a way around these errors or any reasons
>>> as to why?
>>> I suspect a low SNR as well. Maybe a problem in the normalisation?
>>> Thanks,
>>> Ludo
>>>
>>>
>>> On Mon, Oct 14, 2013 at 5:39 PM, Amedee des Georges
>>> <[log in to unmask] <mailto:[log in to unmask]>> wrote:
>>>
>>> I tried removing micrographs with small defoci, but still
>>> get the error below:
>>>
>>> Expectation iteration 3 of 50
>>> 000/??? sec ~~(,_,"> [oo]
>>> exp_thisparticle_sumweight= nan
>>> exp_thisparticle_sumweight= nan
>>> exp_thisparticle_sumweight= nan
>>> exp_thisparticle_sumweight= nan
>>> exp_thisparticle_sumweight= nan
>>> exp_thisparticle_sumweight= nan
>>> exp_thisparticle_sumweight= nan
>>> exp_thisparticle_sumweight= nan
>>> exp_thisparticle_sumweight= nan
>>> exp_thisparticle_sumweight= nan
>>> exp_part_id= 111316exp_iimage=1
>>> group_id= 1765 mymodel.scale_correction[group_id]= nan
>>> exp_ipass= 0
>>> sampling.NrDirections(0, true)= 48 sampling.NrDirections(0,
>>> false)= 48
>>> sampling.NrPsiSamplings(0, true)= 12
>>> sampling.NrPsiSamplings(0, false)= 12
>>> mymodel.sigma2_noise[exp_ipart]=
>>> 0.00015
>>> 4.7e-05
>>> 4.7e-05
>>> 3.9e-05
>>> 3.7e-05
>>> .
>>> .
>>> .
>>> 0
>>> 0
>>> 0
>>> exp_thisparticle_sumweight= 0
>>> nan 0
>>>
>>> 0
>>> 0
>>> 0
>>> 0
>>> 0
>>> 0
>>> 0
>>> 0
>>> 0
>>>
>>> mymodel.avg_norm_correction= 0.671374
>>> wsum_model.avg_norm_correction= 0
>>> written out Mweight.spi
>>> exp_thisparticle_sumweight= nan
>>> exp_min_diff2[exp_ipart]= 9.9e+100
>>> slave 40 encountered error: ERROR!!! zero sum of weights....
>>> File: src/ml_optimiser.cpp line: 3938
>>> +++ RELION: command line arguments (with defaults for
>>> optional ones between parantheses) +++
>>>
>>>
>>>
>>>
>>> --
>>> Amedee des Georges
>>> Frank lab
>>> Howard Hughes Medical Institute
>>> Dept. of Biochemistry and Molecular Biophysics
>>> 650 W 168th St, P&S Black Building room 221
>>> New York, NY 10032
>>> (T) 212 305-9527
>>> (F) 212 305-9500
>>> (E) [log in to unmask] <mailto:[log in to unmask]>
>>> http://franklab.cpmc.columbia.edu
>>> <http://franklab.cpmc.columbia.edu/>
>>>
>>>
>>>
>>> On Oct 4, 2013, at 2:34 PM, Amedee des Georges
>>> <[log in to unmask] <mailto:[log in to unmask]>> wrote:
>>>
>>> > Here is what I get after changing
>>> >
>>> > mymodel.scale_correction[igroup] = 0.;
>>> > to:
>>> > mymodel.scale_correction[igroup] = 1.;
>>> >
>>> > WARNING: ignoring group 845 with very small or negative
>>> scale (6.91088e-310); Use larger groups for more stable
>>> scale estimates.
>>> > WARNING: ignoring group 2035 with very small or negative
>>> scale (6.94752e-310); Use larger groups for more stable
>>> scale estimates.
>>> > WARNING: ignoring group 1094 with very small or negative
>>> scale (6.94713e-310); Use larger groups for more stable
>>> scale estimates.
>>> > WARNING: ignoring group 2403 with very small or negative
>>> scale (6.93112e-310); Use larger groups for more stable
>>> scale estimates.
>>> >
>>> > It seem to be doing this for the whole dataset.
>>> >
>>> >
>>> >
>>> > On Oct 4, 2013, at 10:58 AM, Sjors Scheres
>>> <[log in to unmask]
>>> <mailto:[log in to unmask]>> wrote:
>>> >
>>> >> Hi Amedee,
>>> >> Did you do this?
>>> >> http://www2.mrc-lmb.cam.ac.uk/relion/index.php/Known_issue
>>> >> S
>>> >>
>>> >> On 10/04/2013 03:07 PM, Amedee des Georges wrote:
>>> >>> As I feared, the problem persists with the official
>>> release. I suspect some of the micrographs to have very low
>>> SNR and trigger this bug.
>>> >>>
>>> >>>
>>> >>>
>>> >>> Expectation iteration 2 of 50
>>> >>> 52.70/52.70 min
>>> ............................................................~~(,_,">
>>> >>> Maximization ...
>>> >>> 9/ 9 sec
>>> ............................................................~~(,_,">
>>> >>> Estimating accuracies in the orientational assignment ...
>>> >>> 1.42/1.42 min
>>> ............................................................~~(,_,">
>>> >>> Auto-refine: Estimated accuracy angles= 23.18 degrees;
>>> offsets= 5.775 pixels
>>> >>> Auto-refine: WARNING: The angular accuracy is worse than
>>> 10 degrees, so basically you cannot align your particles (yet)!
>>> >>> Auto-refine: WARNING: You probably need not worry if the
>>> accuracy improves during the next few iterations.
>>> >>> Auto-refine: WARNING: However, if the problem persists
>>> it may lead to spurious FSC curves, so be wary of inflated
>>> resolution estimates...
>>> >>> Auto-refine: WARNING: Sometimes it is better to tune
>>> resolution yourself by adjusting T in a 3D-classification
>>> with a single class.
>>> >>> Estimated required memory for expectation step= 0.384728
>>> Gb, maximum allowed memory = 8 Gb.
>>> >>> CurrentResolution= 31.688 Angstroms, which requires
>>> orientationSampling of at least 11.25 degrees for a particle
>>> of diameter 320 Angstroms
>>> >>> Oversampling= 0 NrHiddenVariableSamplingPoints= 74880
>>> >>> OrientationalSampling= 30 NrOrientations= 576
>>> >>> TranslationalSampling= 6 NrTranslations= 13
>>> >>> =============================
>>> >>> Oversampling= 1 NrHiddenVariableSamplingPoints= 2396160
>>> >>> OrientationalSampling= 15 NrOrientations= 4608
>>> >>> TranslationalSampling= 3 NrTranslations= 52
>>> >>> =============================
>>> >>> Expectation iteration 3 of 50
>>> >>> 000/??? sec ~~(,_,"> [oo]
>>> exp_thisparticle_sumweight= nan
>>> >>> exp_thisparticle_sumweight= nan
>>> >>> exp_thisparticle_sumweight= nan
>>> >>> exp_thisparticle_sumweight= nan
>>> >>> exp_thisparticle_sumweight= nan
>>> >>> exp_thisparticle_sumweight= nan
>>> >>> exp_thisparticle_sumweight= nan
>>> >>> exp_thisparticle_sumweight= nan
>>> >>> exp_thisparticle_sumweight= nan
>>> >>> exp_thisparticle_sumweight= nan
>>> >>> exp_thisparticle_sumweight= nan
>>> >>>
>>> >>> On Oct 3, 2013, at 3:12 AM, Sjors Scheres
>>> <[log in to unmask]
>>> <mailto:[log in to unmask]>> wrote:
>>> >>>
>>> >>>> Dear Amedee,
>>> >>>> Please first update to the official relion-1.2 release.
>>> The beta-releases
>>> >>>> may well contain bugs.
>>> >>>> best, S
>>> >>>>
>>> >>>>>
>>> >>>>> Hi,
>>> >>>>>
>>> >>>>> I have an unusual error at the 3rd iteration of
>>> classification on a new
>>> >>>>> dataset. This happens as the resolution dropped from
>>> 80 to 30A at the
>>> >>>>> previous iteration.
>>> >>>>>
>>> >>>>> Expectation iteration 3 of 50
>>> >>>>> 000/??? sec ~~(,_,">
>>> >>>>> [oo] exp_thisparticle_sumweight= nan
>>> >>>>> exp_thisparticle_sumweight= nan
>>> >>>>> exp_thisparticle_sumweight= nan
>>> >>>>> exp_thisparticle_sumweight= nan
>>> >>>>> exp_thisparticle_sumweight= nan
>>> >>>>> exp_thisparticle_sumweight= nan
>>> >>>>> exp_thisparticle_sumweight= nan
>>> >>>>>
>>> >>>>> I didn't manage to fix, so any help is welcome. This
>>> is with relion 1.2b7,
>>> >>>>> which was running fine until now.
>>> >>>>>
>>> >>>>> Amedee
>>> >>>>>
>>> >>>>>
>>> >>>>
>>> >>>> --
>>> >>>> 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
>>> >>
>>> >>
>>> >
>>>
>>>
>>
>
>
--
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
|