Hi Grigory,
What happens if you put that threshold a bit higher, e.g. 0.85 or 0.9?
Your FSC drops just at around 0.8 (which is strange!). Also, your
corrected curves look a bit strange. Is your mask perhaps too tight? Are
the FSC values for the phase-randomised map around 0 at the estimated
resolution after correction? You should try and make you mask soft and
featureless enough to make that happen. You can also use
relion_mask_create from the command-line and input a low-pass filtered
reconstruction (e.g. to 15A) to create your mask. That way, the mask may
be 'smoother' and thereby have less convolution effects.
HTH,
Sjors
> Hello Sjors,
>
> it looks like the spike is still present in relion-1.4. I have attached
> the
> FSC curves before and after post-processing. I also played
> with --randomize_at_fsc and the spike is moving.
>
>
>
> Best regards,
> Grigory
>
> --------------------------------------------------------------------------------
> Grigory Sharov, Ph.D.
> Institute of Genetics and Molecular and Cellular Biology
> Integrated Structural Biology Department (CBI)
> 1, rue Laurent Fries
> 67404 Illkirch, France
> tel. 03 69 48 51 00
> e-mail: [log in to unmask]
>
> On Thu, Oct 8, 2015 at 6:21 PM, Sjors Scheres <[log in to unmask]>
> wrote:
>
>> Thanks for digging up that old message, Ludo!
>> Just to add that in relion-1.4, this dip is now removed from the curve.
>> HTH,
>> Sjors
>>
>>
>> On 10/08/2015 02:28 PM, Ludovic Renault wrote:
>>
>>> Hi Miao,
>>>
>>> That's a point that's been raised before.
>>> I copy below the prevous messages from Sjors.
>>>
>>> Best,
>>> Ludo
>>>
>>> --
>>> Hi Jianhua,
>>> We see this artefact also sometimes, but you do not need to worry about
>>> it. Probably it has to do with correlations "spilling over" into the
>>> next
>>> (1 or sometimes 2) resolution shell(s) after the phase-randomization,
>>> leading thereby to an over-correction. We sometimes just omit that
>>> point
>>> from the corrected curve. Perhaps I'll do something along those lines
>>> in
>>> the next release. You should of course be very suspicious if the entire
>>> corrected curve drops rapidly after the phase randomization, but just a
>>> dip of 1 (or 2) shell(s) is OK. In your case, I would be more worried
>>> about the corrected curve having a plateau that is much lower than the
>>> near-1 value just after this dip. Are you sure your mask is not too
>>> tight?
>>> We often put a soft edge of up to 10 Angstroms on our masks.
>>> HTH,
>>> S
>>> --
>>>
>>> Jianhua,
>>> The corrected FSC curve usually does not have a plateau that is much
>>> lower
>>> than before the phase randomisation. A too tight mask that cuts protein
>>> density or that has a too sharp edge will lead to over-correction. The
>>> dip
>>> is probably caused by correlations in the signal because even without
>>> masking the molecule lives only in the center of an otherwise empty
>>> box.
>>> HTH,
>>> S
>>>
>>> On Thu, Oct 8, 2015 at 1:57 PM, Miao Gui <[log in to unmask]> wrote:
>>>
>>> Dear all,
>>>>
>>>> After Auto-refine, I do Post-process in Relion1.3 and get a fsc curve
>>>> shown as the attached file.
>>>> There is a sharp spike at ~16Å resolution. I'm just wondering if this
>>>> is
>>>> a
>>>> normal phenomenon, or something goes wrong.
>>>> I have tried auto-B factor estimation and my own B-factor, the fsc
>>>> curves
>>>> are alike. And when I fit a related crystal structure with my EM map,
>>>> it
>>>> looks pretty well.
>>>>
>>>> Best regards,
>>>>
>>>> Miao
>>>>
>>>>
>> --
>> 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
|