[phenixbb] The 5'end guanosine triphosphate is not linked to the rest of RNA model by Phenix refine

7 Jul 2020

      Dear Phenix Community,

I'm at beginner level of using Phenix. I have difficulty to refine a protein-RNA complex in Phenix.

I refined the protein model without RNA first, then I built the RNA model into the difference density map in Coot through Calculate>Other Modelling Tools>Ideal DNA/RNA, where I input the sequence and fit nucleotides into the density. At the 5' end it's guanosine triphosphate (GTP) but the modeler tool in Coot only generates nucleoside monophosphate. So I built the RNA model from the second nucleotide at 5' end downstream until where RNA density was too poor to model. I did the real space refine zone in Coot to fit this main part of the RNA into its density the best possibly.

For the GTP at 5' end, I realized there is GTP.cif restraint file in Phenix monomer library. So I obtained an arbitrary pdb file for GTP using eLBOW and corrected atom names matching the definition in GTP.cif file. Then I can also fit GTP into the RNA density at 5' end using real space refine zone in Coot. Now I have a broken RNA model with GTP unlinked to the ribose-phosphate backbone.

For Phenix refinement, I supplied a cif_link_GTP-G.params file to define the linkage for the two RNA chains in the complex:

refinement.pdb_interpretation {
  apply_cif_link {
    data_link = p
    residue_selection_1 = chain C and resname GTP and resid 1
    residue_selection_2 = chain C and resname G and resid 2
  }
  apply_cif_link {
    data_link = p
    residue_selection_1 = chain D and resname GTP and resid 1
    residue_selection_2 = chain D and resname G and resid 2
  }
}

After refinement, I noticed O3' of GTP appears connected to P of the following nucleotide G in the refined model in Coot. However, if I do the real space refine zone in Coot, the linkage becomes broken, i.e., I cannot refine the RNA model as a whole: Coot only allows separate real space refine for GTP and the rest. It seems Coot still cannot recognize GTP and the rest as one body. In the refinement settings there is a button for Automatic linking options.  Interestingly, I noticed if I leave the automatic covalent linking as default, Phenix writes Link records in the output pdb after refinement. But it also links one of the protein residues to the GTP in chain D. In order to avoid the unwanted intermolecular linkage, I clicked Link none for Automatic linking options, then there is no any link records written in the output pdb despite of the defined linkage in cif_link_GTP-G.params file. So it seems for me Phenix ignores the input cif_link_GTP-G.params file, it writes Link records only if the Automatic linking option is activated. I worried whether it could be due to the syntax of the cif_link_GTP-G.params file. So I tried the following syntax in two separate files, the same issue was observed.

cif_link_GTP-G_chC.params
refinement.pdb_interpretation.apply_cif_link {
    data_link = p
    residue_selection_1 = chain C and resname GTP and resid 1
    residue_selection_2 = chain C and resname G and resid 2
}

cif_link_GTP-G_chD.params
refinement.pdb_interpretation.apply_cif_link {
    data_link = p
    residue_selection_1 = chain D and resname GTP and resid 1
    residue_selection_2 = chain D and resname G and resid 2
}

I also need to point out that p linkage is pre-defined in the list of links (mon_lib_list.cif) in Phenix monomer library.

I apologize for this long explanation of the problem. I hope it is clear.

I really appreciate your advice in advance.

Sincerely,
Zixian Li