Bio3DEM

标题: postprocessing problem [打印本页]

作者: zzh8084 时间: 2017-7-21 17:12
标题: postprocessing problem

   大家好，我有个Postprocessing problem，3D Classification计算已经到14埃了，但接着做Postprocessing，出现了WARNING，并提示如下，不知是哪里出现了问题，求解( ⊙ o ⊙ )啊！O(∩_∩)O谢谢
== Masking input maps ...
  + randomize phases beyond:    20 Angstroms
WARNING: The phase-randomised FSC is larger than 0.10 at the estimated resolution!
WARNING: This may result in an incorrect resolution estimation. Provide a softer mask with less features to get lower phase-randomised FSCs

3D Classification：_rlnCurrentResolution                   14.000000

chimera threshold：0.0037

== Reading input half-reconstructions:
  + half1-map:                   D1/run_2D_sel_clx_cls3_D1_3DsM_ct20_half1_class001_unfil.mrc
  + half2-map:                   D1/run_2D_sel_clx_cls3_D1_3DsM_ct20_half2_class001_unfil.mrc
== Perform auto-masking ...
  + density threshold:          0.00405
  + extend ini mask:             5 pixels
  + width soft edge:             5 pixels
== Extending initial binary mask ...
1.03/1.03 min ............................................................~~(,_,">
== Making a soft edge on the extended mask ...
  51/  51 sec ............................................................~~(,_,">
== Masking input maps ...
  + randomize phases beyond:    20 Angstroms
WARNING: The phase-randomised FSC is larger than 0.10 at the estimated resolution!
WARNING: This may result in an incorrect resolution estimation. Provide a softer mask with less features to get lower phase-randomised FSCs.
== Dividing map by the MTF of the detector ...
  + mtf STAR-file:                mtf-k2-200kev.star
== Applying sqrt(2*FSC/(FSC+1)) weighting (as in Rosenthal & Henderson, 2003) ...
== Fitting straight line through Guinier plot to find B-factor ...
  + fit from resolution:          10
  + fit until resolution:       0
  + slope of fit:                -135.631
  + intercept of fit:             -14.1714
  + correlation of fit:          0.540263
  + apply b-factor of:          -542.524
== Low-pass filtering final map ...
  + filter frequency:             12
== Writing out put files ...
  + Processed map:                D1/run_2D_sel_clx_cls3_D1_3DsM_ct20_p2.mrc
  + Processed masked map:       D1/run_2D_sel_clx_cls3_D1_3DsM_ct20_p2_masked.mrc
  + Auto-mask:                   D1/run_2D_sel_clx_cls3_D1_3DsM_ct20_p2_automask.mrc
  + Metadata file:                D1/run_2D_sel_clx_cls3_D1_3DsM_ct20_p2.star
  + FINAL RESOLUTION:             12

作者: zzh8084 时间: 2017-7-21 17:19
@小雨

作者: 小雨 时间: 2017-7-22 16:48
你好呀zzh~

Relion的post processing在计算FSC的时候，会加入mask。
为了防止这个mask太紧，过高的估计分辨率，relion会自动生成一个噪音的对照组，加上同样的mask，如果这个mask使得噪音也产生了大于0.10的FSC，就说明你所估计的分辨率，除了信号的相关性之外，还有你自己的mask的相关性。这个时候relion就会warning你~

解决办法是把mask搞得松一些，改变这三个值：
== Perform auto-masking ...
  + density threshold:          0.00405
  + extend ini mask:             5 pixels
  + width soft edge:             5 pixels

作者: zzh8084 时间: 2017-7-22 21:30

小雨发表于 2017-7-22 16:48
你好呀zzh~

Relion的post processing在计算FSC的时候，会加入mask。

非常感谢！！

作者: Zheng_Liu 时间: 2017-8-3 11:24

zzh8084 发表于 2017-7-22 21:30
非常感谢！！

记得给有营养的帖子,评分哦.

作者: zzh8084 时间: 2017-8-3 12:28

Zheng_Liu 发表于 2017-8-3 11:24
记得给有营养的帖子,评分哦.

好的，

欢迎光临 Bio3DEM (http://bio3dem.com/)