I2PC / scipion

Scipion is an image processing framework to obtain 3D models of macromolecular complexes using Electron Microscopy (3DEM)
http://scipion.i2pc.es
Other
76 stars 47 forks source link

Store the information about particles distribution after gold standard #1172

Open 1aviervargas opened 7 years ago

1aviervargas commented 7 years ago

Hi there,

I am working on the enrich method. After discussion with Coss and JM we decided to use gold standard somehow to show we have not model bias problems. The point is that I need the information about how particles are distributed between the two halves after autorefine. In Scipion this information is not stored. I have edited relion import to incorporate this information (tested) and also the relion refine protocol (not tested yet); @delarosatrevin and @pconesa, can you take a look to check that this takes sense? The branch is jv_hetero_enrich

The edited files are

relion/convert.py

if partRow.hasLabel(md.RLN_PARTICLE_RANDOM_SUBSET):
    img._random_Subset = Integer(partRow.getValue(md.RLN_PARTICLE_RANDOM_SUBSET))

relion/protocol_refine3D.oy

    item._random_Subset = Integer(row.getValue(md.RLN_PARTICLE_RANDOM_SUBSET))

Thanks

pconesa commented 7 years ago

I'll assign this to @josuegbl, he can better check this than me.

josuegbl commented 7 years ago

@1aviervargas, did you submitted the proposed changes? I think you should add _rln before the name of the label, cause this label is specific to Relion. Maybe you can consider split your set into two subsets randomly in your protocol, instead storing a relion label.

1aviervargas commented 7 years ago

Hi @josuegbl https://github.com/josuegbl I did in my branch but not merged with debel yet. Should I merge only this change? In addition, the problem I see with what you suggested is that if any other protocol wants to expose this information this label is not adequate. I agree that maybe is better to add the label _rlnHalfId now and then think if we include this information in Scipion model after, agree?

On Thu, Aug 17, 2017 at 5:16 AM, Josué Gómez Blanco < notifications@github.com> wrote:

@1aviervargas https://github.com/1aviervargas, did you submitted the proposed changes? I think you should add _rln before the name of the label, cause this label is specific to Relion. Maybe you can consider split your set into two subsets randomly in your protocol, instead storing a relion label.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/I2PC/scipion/issues/1172#issuecomment-323015017, or mute the thread https://github.com/notifications/unsubscribe-auth/ALs-YZMFE_tGiGEzwUmMOPNVaa9XZ0q8ks5sZATygaJpZM4O0Aem .

azazellochg commented 4 years ago

While relion stores the half id, afaik cryosparc does the split randomly, so I'm not sure how to go with this between plugins