Hi,
Recently I submitted the main map and the half maps from the 3D flex reconstruction job to EMDB. Yesterday, I received a response from them to submit unmasked, unfiltered, and otherwise unprocessed half-maps. What should I do in this case? I mean how to get unmasked half-maps from 3D-flex reconstruction job? (Particles and volume I used for running 3D-flex job come from the NU-refinement job.)
Does anyone have submitted a map from 3D flex job to EMDB? Which half-map did you use? Half maps from parent job (NU-refinement) or did you re-run another refinement using inputs from 3D flex job to get unmasked half-maps?
P.S. 3D flex improved the quality and resolution of the Non-Uniform-refined map from 3.65 to 3.39A, this is the reason I am pushing to deposit the map from 3D flex.
Regards,
Shashank.
Hi @Shashank,
Great to hear that 3DFlex improved the quality of your structure!
In terms of half-maps, 3DFlex reconstruct job creates two half-maps that are independent (i.e. gold-standard split) but due to the way the reconstruction works (it’s done in real-space using an LBFGS optimizer to be able to optimize under the motion model, rather than in Fourier-space using backprojection which assumes rigidity of the particle), the half-maps are only solved within the region where the outer solvent mask that was provided is greater than zero. When these half-maps are used to compute the FSC, the soft mask that is used is contained within the solvent region, so it doesn’t matter that the half-maps are already zero outside the solvent region (since the soft mask would be zero outside the solvent region anyway).
In terms of deposition, I’m not sure if you can explain the above to the people at EMDB or if the response you got was an automated response. It would be great if you can try to explain - that would be the simplest way forward. At the moment I can’t think of a straightforward way to get half-maps from 3DFlex that also contain noise in the solvent region (instead of being zero).