Hi,
Currently, the new ab initio refinement job generates and aligns the two half maps, but does not perform reconstruction. It might be useful as a final step, if ab initio refinement performed a homogeneous reconstruction (so the user can judge the resolution, and proceed directly to local refinement). Currently I do this as a separate job, but it would be convenient to include it I think.
Also (longer term), I wonder if it would be possible/worthwhile to incorporate periodic FSC filtering during ab initio refinement, perhaps every epoch, now that half sets are available? Presumably the FSC-filtered (or perhaps NU regularized?) half map should be a better estimate of the true structure than the unfiltered map, so I wonder if this would help in difficult cases?
Cheers
Oli
