Hi,
Thanks all for the input. Indeed I ran per-particle B-factor fitting. Will turn off the fitting and report back.
Best,
Minglei
> On Dec 4, 2019, at 10:45 AM, Takanori Nakane <[log in to unmask]> wrote:
>
> Hi,
>
> Probably your B factor refinement was not successful.
> (It does not necessarily work nicely as I explained in the previous mail)
>
> Please use STAR files before B factor refinement or
> remove rlnCtfBfactor and rlnCtfScalefactor columns.
>
> Best regards,
>
> Takanori Nakane
>
>> Hi,
>>
>> ??? I see a problem that is possibly related to the one just reported
>> by several people: Refinement in relion-3.1-beta runs nicely but upon
>> combining the two half maps it reports a resolution of 0.871938 A, which
>> is obviously impossible. Nevertheless, postprocessing works fine giving
>> me realistic resolution but the FSC curve is messed up similarly as
>> reported by Minglei; it goes up again at the right side. I noticed that
>> in the last iteration several warnings were issued like:
>>
>> ? 112590 in group 13; Are your groups large enough? Or is the reference
>> on the correct greyscale?
>> 2.19/2.19 hrs
>> ............................................................~~(,_,">
>>
>> I also noticed that some values of? _rlnLogLikeliContribution #23 were
>> negative in the final run_data.star!
>>
>> Nevertheless, the final map looks good!
>>
>> best, Dieter
>>
>> ------------------------------------------------------------------------
>> Dieter Blaas
>> Max Perutz Laboratories
>> Medical University of Vienna,
>> Inst. Med. Biochem., Vienna Biocenter (VBC),
>> Dr. Bohr Gasse 9/3,
>> A-1030 Vienna, Austria,
>> Tel: 0043 1 4277 61630,
>> Mobile: 0043 699 1942 1659
>> Fax: 0043 1 4277 9616,
>> e-mail: [log in to unmask]
>> ------------------------------------------------------------------------
>>
>> Am 04.12.2019 um 13:53 schrieb Takanori Nakane:
>>> Hi,
>>>
>>> Indeed B factor refinement is still new and we don't know when
>>> it is beneficial. I recommend you to leave it OFF during processing
>>> and only try it at the very end to see if it improves the resolution.
>>>
>>> You can also try these options in relion_ctf_refine to avoid
>>> excessive B factors.
>>>
>>> --bfac_min_B (-30) : Minimal allowed B-factor
>>> --bfac_max_B (300) : Maximal allowed B-factor
>>>
>>> For example, you can make the range smaller (say, 0 to 150).
>>>
>>> Best regards,
>>>
>>> Takanori Nakane
>>>
>>>> FYI I've seen a similar effect at FSC high-frequencies when using
>>>> B-factor refinement in relion v3.1
>>>>
>>>> Cheers,
>>>> Joe
>>>>
>>>> On 03/12/2019, 21:08, Minglei Zhao wrote:
>>>>> OK. Will report back.
>>>>>
>>>>> Thanks!
>>>>>
>>>>> Minglei
>>>>>
>>>>>> On Dec 3, 2019, at 2:50 PM, Takanori Nakane
>>>>>> <[log in to unmask]>
>>>>>> wrote:
>>>>>>
>>>>>> Hi,
>>>>>>
>>>>>> Interesting.
>>>>>>
>>>>>>> After CtfRefine (first beam tilt, then anisotropic magnification,
>>>>>>> then
>>>>>>> defocus)
>>>>>> Can you run Refine3D on outputs from each step to see which one was
>>>>>> harmful?
>>>>>>
>>>>>> Best regards,
>>>>>>
>>>>>> Takanori Nakane
>>>>>>
>>>>>> On 2019/12/03 20:33, Minglei Zhao wrote:
>>>>>>> Just double checked, no intermediate half maps were used in any post
>>>>>>> processing.
>>>>>>> Minglei
>>>>>>>> On Dec 3, 2019, at 2:30 PM, Takanori Nakane
>>>>>>>> <[log in to unmask]> wrote:
>>>>>>>>
>>>>>>>> Hi,
>>>>>>>>
>>>>>>>> First of all, is "Relion 3.0 beta" typo of
>>>>>>>> RELION 3.1 beta? Otherwise, you should use at least RELION 3.0.8.
>>>>>>>>
>>>>>>>>> the high frequency signal is abnormal (why?), but the curve still
>>>>>>>>> looks
>>>>>>>> Probably you used half maps from intermediate iterations in
>>>>>>>> PostProcess.
>>>>>>>> Please use the converged half maps and repeat subsequent steps.
>>>>>>>>
>>>>>>>> Best regards,
>>>>>>>>
>>>>>>>> Takanori Nakane
>>>>>>>>
>>>>>>>> On 2019/12/03 20:25, Minglei Zhao wrote:
>>>>>>>>> Dear Microscopists,
>>>>>>>>> I encountered a weird behavior of FSC for an apoferritin dataset
>>>>>>>>> processed using Relion 3.0 beta:
>>>>>>>>> After 3D refinement and post process, everything looks good:
>>>>>>>>> After CtfRefine (first beam tilt, then anisotropic magnification,
>>>>>>>>> then defocus), refined again, and post process, the high frequency
>>>>>>>>> signal is abnormal (why?), but the curve still looks OK before
>>>>>>>>> 0.7(1/A). Estimated resolution improved a little bit:
>>>>>>>>> After Bayesian Polishing, refined again using mask and
>>>>>>>>> solvent-flattened FSC, the estimated resolution is 1.9 A, FSC from
>>>>>>>>> Refine3D job looks fine:
>>>>>>>>> However, using the same mask in the previous refinement to filter
>>>>>>>>> and sharpen the map, the FSC curve is messed up:
>>>>>>>>> Any obvious things I did wrong?
>>>>>>>>> Thanks for your attention.
>>>>>>>>> Minglei
>>>>>>>>> ------------------------------------------------------------------------
>>>>>>>>> To unsubscribe from the CCPEM list, click the following link:
>>>>>>>>> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPEM&A=1
>>>>> ########################################################################
>>>>>
>>>>> To unsubscribe from the CCPEM list, click the following link:
>>>>> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPEM&A=1
>>>> ########################################################################
>>>>
>>>> To unsubscribe from the CCPEM list, click the following link:
>>>> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPEM&A=1
>>>>
>>> ########################################################################
>>>
>>> To unsubscribe from the CCPEM list, click the following link:
>>> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPEM&A=1
>>
>> ########################################################################
>>
>> To unsubscribe from the CCPEM list, click the following link:
>> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPEM&A=1
>>
>
> ########################################################################
>
> To unsubscribe from the CCPEM list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPEM&A=1
########################################################################
To unsubscribe from the CCPEM list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPEM&A=1
|