Hello all
I am currently in the process of reconstructing a very large composite map, for deposition with a very large model (49 unique chains i believe). The model constitutes 1/3 of a C3/C6 structure, and the consensus map covers the core of this structure nicely. However, the periphery was poorly resolved but could be easily interpreted through 2 local refinements. While the central part of the density is C3, the periphery is C6. The central C3 structure is adapted to the outer C6 structure via adaptor proteins that are fully resolved in the consensus map, such that by the time the density reaches the locally refined regions, the density appears truly C6.
The local maps were generated by using C6 symmetry expansion on the C3 core map, followed by a box shift to the location, reconstruction without alignment, and local refinement of the reconstruction. These maps then integrate seamlessly into the core map using chimeraX fit in map command.
I am however, hitting an issue with how to deposit this.
For modelling, i have used phenix combine_focussed_maps to generate a composite of a C3 asymmetric unit only by cropping the various maps; this includes the central proteins modelled into their slight conformational variations that are adopted to facilitate adaption to the outer C6 region, as well as two copies of each of the C6 proteins (e.g. proteins with 6 copies that are C3 have been modelled into two conformational variants, proteins with 6 copies that are C6 have been duplicated).
This works well for modelling, however, I figure there may be an issue for deposition. I feel its probably unwise to deposit a cropped composite map? is this correct?
The alternative is to produce a composite map using 6 copies of each local map duplicated into symmetry positions on the consensus map; however i strait up cannot run this using phenix, its just too large.
The second option i am aware of is volume maximum command in chimeraX. However, I have found that regions where local maps becoming highly blurred will often dominate the good regions of your concensus map using this command, resulting in noisy transitional regions between the maps.
A solution i have previously used purely for interpretation, is to crop the local and consensus maps manually using the models as a guide to the point where the density starts to lose local resolution (normally with some additional padding). There is a nice level of quality overlap between my maps, so if i do this then merging the maps omits the noise problem producing seamless density. This is not altogether different from how phenix works from my understanding, which weights the density values based on the quality of local model fit.
My second question is then:
Is it okay to deposit a composite map that i have manually truncated to facilitate this kind of merging? I am at a bit of a loss of how to deposit this monster without this technique! using that method i can combine the 2 x 6 local maps + consensus map into a high-resolution composite with seamless density, which is otherwise a computational bottleneck using phenix.
Sorry for the length of this post, but I hope you can appreciate the complexity and I wanted to be honest about all I have done!
Cheers,
James.