Closed nelse003 closed 5 years ago
Email from Nick, regarding refinement with multiple input files
The problem is not with pandda.export but with XCE and that there is no ability to add additional CIF files for a structure beyond those of the soaked ligand.
I can tell you the source of the problem, which might enable you to make a workaround, and will provide a starting point for TobRachEllio to program an XCE solution.
The problem: ONLY ONE CIF FILE CAN BE PROVIDED TO REFMAC.
For refmac to refine a structure with two or more ligands, the individual cif files have to be merged into one file.
You can do that with a VERY archaic program called LIBCHECK from CCP4.
LIBCHECK <<eof _DOC N _FILE_L file1.cif _FILE_L2 file2.cif _FILE_O out_prefix _END eof
which will create a file called out_prefix.lib, which you can rename to out_prefix.cif.
You’ll have to do this for each ligand in each dataset
Dataset1: main.cif + fragment1.cif > combined1.cif Dataset2: main.cif + fragment2.cif > combined2.cif Dataset3: main.cif + fragment3.cif > combined3.cif Dataset4: main.cif + fragment4.cif > combined4.cif
which of course is easily scriptable…
a) Pretty sure you can simply concatenate all ligand cif files (that's the whole point of the CIF format.)
b) co-factor restraints already exist in the CCP4 monomer library: it would be very rare to need to generate from scratch.
This feels very automatable: any .cif files in the same place as the reference pdb simply get strung onto the end of the fragment.cif file.
a) Pretty sure you can simply concatenate all ligand cif files (that's the whole point of the CIF format.)
in general it's not that straightforward, link IDs must be unique so if the same IDs are used in two files they must be renamed.
A recurring issue amongst users: Users can already provide additonal CIF files during initial refinement with DIMPLE. However, it is not necessarily obvious to them how XCE processes the infomation. It was suggested to present the presence of a suitable CIF file (same name as PDB file) during initial refinement by showing it in the combobox, e.g. model1.pdb/ model1.cif model2.pdb/ NO CIF model3.pdb/ model3.cif This will of course only help during initial refinement. What should then happen is:
implemented in v1.3.8.12 (10/09/2019)
Currently any co-factor restraints need to be supplied by adjusting the cif file of the bound ligand. This is awkward as it cannot be done within the xce interface, and is inconsitent with crystallogrpahic protocol.
Solution: Ability for user to supply/ generate a/multiple cif file for any co-factors defined. As option in xce interface.