Streaky artifacts on post-processed map

[William Nicolas]

Hello Yuntao,

I tested spIsoNet on a dataset that suffered anisotropy to see if it can correct it.

The data comes from cryosparc. I exported the two half-maps and the solvent mask. This is the gFSC curve from cryosparc:

I succesfully created the corrected half maps using spisonet.py reconstruct. I checked them in ChimeraX and they look OK.

I now want to joing them in Relion as indicated in your tutorial. I used the following parameters: 

`which relion_postprocess` --mask cryosparc_P50_J180_007_volume_mask_refine.mrc --i isonet_maps/run_half1_corrected.mrc --o PostProcess/job005/postprocess  --angpix 0.83 --mtf ../../repo/relion/data/mtf_falcon4EC_300kV.star --mtf_angpix 0.83 --auto_bfac  --autob_lowres 6  --pipeline_control PostProcess/job005/

It goes through but I get the following warning: "the unmasked FSC extends beyond the solvent-corrected FSC. Skip masking for now, but you may want to adjust you mask

You can force the mask by the --force_mask option"

The postprocessing.mrc map shows streaks and lots of noise:

And the FSC and Guinier curves (attached) also look weird. Is this because of the mask issue?

Am I doing something wrong?

Thank you,

[YUNTAO LIU]

  Hi William,

If the half maps look fine, you probably need to specify --limit_res and also try different B factors in relion post processing.

The FSC estimation for spIsoNet corrected half maps is no longer gold standard (unless you specify —independent). So that the automatic post processing will not find the correct resolution and b factor.

--
You received this message because you are subscribed to the Google Groups "spIsoNet" group.
To unsubscribe from this group and stop receiving emails from it, send an email to spisonet+u...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/spisonet/be7438b4-a927-477e-a867-af957b7d15a5n%40googlegroups.com.

--

Best Regards,

Yuntao Liu,  Postdoc.

California NanoSystem Institute

University of California Los Angeles

[William Nicolas]

Sorry, I think I answered you privately. I'll repeat publicly for the record ;-) 

You are talking about the spisonet.py reconstruct --limit_res option right? If our map goes to 2.8A, what should be the limit? I thought 3.5 was enough.

[YUNTAO LIU]

Hi William,

I  want to say --low_pass in relion post processing. You can limit it to the real resolution it has and also try a different B factor. 

For spIsoNet, the limit to 3.5A should be enough. Sometimes cryosparc overestimates the resolution.

--
You received this message because you are subscribed to the Google Groups "spIsoNet" group.
To unsubscribe from this group and stop receiving emails from it, send an email to spisonet+u...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/spisonet/9b35e13a-2476-4f54-acd8-127923b8a5e6n%40googlegroups.com.

[William Nicolas]

Thanks,

That seemed to work!