FSC curve during Local refinement looks weird

Hi, I was running a local refinement job with dynamic masking and after the job was finished the FSC curve looks weird. I have two questions:

  1. Why the tight mask is not touching zero in the upper FSC curve? Is the mask too tight? Do I need to play with Near/Far options in dynamic masking?

  2. Why there is a dip at around 4Å for the corrected mask in the FSC curve below?

Any suggestion is appreciated.

Hi @sdas,

  1. Yes, this could be because the mask is too tight. You could increase the “Dynamic mask far” parameter to make the edge softer. But the recommended practice is to keep static masking enabled (as it is by default) and simply generate the mask from the initial volume, using Volume Tools. Please see the membrane protein tips guide page, which also links out to the mask generation tutorial for best practices on mask generation. The local refinement job page also covers some of the best practices for this.
  2. The sharp dip is due to phase randomization. You can refer to the following publication which is implemented in all CryoSPARC refinement jobs: High-resolution noise substitution to measure overfitting

Best,
Michael

1 Like