Tutorial 2. Generation of a new ligand description using JLigand

1. Introduction

In the first tutorial you were able to refine you example structure both globally against X-ray data using Refmac and locally against electron density using Coot's real space refinement. This was only possible because both programs knew about geometrical restraints associated with all covalent bonds in the structure. This is because both amino-acid residues and sugar residues present in that particular structure, as well as covalent linkages between monomers are described in CCP4 dictionary. Obviously the dictionary does not include all possible chemical compounds and sometimes it is necessary to generate new dictionary entries describing covalent bonds in a new ligand.

In this tutorial we demonstrate how it can be done using JLigand included in CCP4 suite. For simplicity, we assume a somewhat artificial situation. Suppose we want to analyse a structure downloaded from the PDB. In addition, we are going  to perform several cycles of refinement and examine the electron density - to make sure that the structure of a ligand binding site is trustworthy (see Notes,  1). In this example we'll examine a recently deposited structure with PDB code 4do4. The structure contains a ligand with PDB code DJN, which has not yet been added to the CCP4 monomer library. Therefore, we will need to generate an additional library containing the template restraints for DJN before any refinement.

2. Procedure

(1) The directory "jligand/2_newligand" contains the  tutorial data (files 4do4.mtz and 4do4.pdb, see also Notes, 2). Set up ccp4i project "1_newligand" with this directory as the project directory.

(2) Run Refmac5 with 4do4.pdb and 4do4.mtz. Use default output file names.
        Input files: 4do4.mtz, 4do4.pdb
        Output files: 4do4_refmac1.mtz, 4do4_refmac1.pdb will not be created

Note that the job has quickly finished and is marked as FAILED in the ccp4i job list. Double-click on the job to open result viewer. It will show Refmac log-file because no graphical output has been generated. You can find the following messages in the log-file explaining why Refmac refused to refine the structure (see Notes, 3).
"WARNING : residue: DJN               510  chain:Aj   - is not in the library"
"Refmac_5.7.0032:  New ligand has been encountered. Stopping now"

(3) The drawing below shows the chemistry of the ligand DJN and atom names

The simplest way to generate the description of DJN using JLigand is detailed below. (If you wish, you may also draw the molecular graph from scratch following instructions in JLigand > Help > JLigand Help.)

Launch JLigand (ccp4i > Refinement > Restraint Preparation > Ligands and links with JLigand)

Import DJN form the pdb-file into JLigand
- File > Open > PDB File; select 4do4.pdb
- Select "DJN" in "Choose Ligand Id" window
- Press "Submit" and JLigand will read coordinates of DJN atoms from the pdb-file and suggest a possible chemical graph.

Correct bond orders if necessary
- double-click on a bond
- select required  bond order from a menu in "Edit Bond Type" window

Regularise (Ligand > Regularise > DJN)

Check chiralities of carbon atoms. They should be correct as the ligand has been imported from a structure that was well refined against high resolution data. However, you may try to change one for exercise:
- Double-click on a chiral carbon atom
- Select "make positive" or "make negative" from a menu in "Edit Atom Details" window (as opposed to "keep negative" or "keep positive")
-  Regularise DJN for changes to have effect
- Do not forget to restore the original chirality

Save ligand description as an additional library
- File > Save as Monomer > DJN
- Keep default name for the library file (DJN.cif)

(5) Rerun Refmac job with additional library DJN.cif
- Select Refmac job and press "Rerun job" button in the ccp4i main window;
- In Refmac task interface, populate "LIB in" with DJN.cif
- Change "Do 10 cycles ..." to "Do 1 cycles ..." in the "Refinement parameters"
- "Run" > "Run now"

(6) Try real space refinement in Coot without and with the additional library loaded
- When Refmac job has finished, open result page and launch Coot from "Output Files" section.
- Navigate to one of the two copies of DJN (Draw > Go To Atom > select A510 DJN from pull down menu)
- Click "Real Space Refine Zone" and then click two times on any atom of the ligand. This does not work because the main monomer library does not contain description of DJN
- File > Import CIF-dictionary; select the project "2_newligand" in the menu initially showing "-Current Dir-" ; select DJN.cif
- Try real space refinement again. Now, with the additional library loaded, Coot knows how to refine the DJN ligand.

3. Notes

(1) The electron density maps for most of the PDB structures are readily available from Uppsala Electron Density Server. In practice, you would not normally need any refinement, but only a quick look at the electron density map to make sure that ligand position and conformation are well defined. This can be done in Coot (File > Fetch PDB & Map using EDS). But, as you may expect, real space refinement would not work for DJN unless you "Import CIF dictionary" DJN.cif.

(2) The file 4do4.pdb was downloaded from the PDB. The file 4do4-sf.cif was downloaded from the PDB and converted to 4do4.mtz using the task "Convert to/ modify/ extend MTZ" located in ccp4i "Reflection Data Utilities" folder.

(3) Actually, DJN happened to be a different PDB code for the compound, which already exists in the CCP4 monomer library under the name NOK. Refmac will figure this out and continue refinement if you select "Check only ligands against Refmac's dictionary description" in the line starting from "Checking against dictionary:", in the folder "Setup Geometric Restraints".