Observant users of HIC-Up sometimes discover that a certain compound is not complete, e.g. "the example of the HEPES that is featured on HIC-Up is missing the hydroxyethyl moeity". This sometimes happens as the HIC-Up-generating script doesn't know any chemistry - it simply takes the highest-resolution copy of every hetero compound and hopes it is complete. The onus is always on the user to assess if the result is relevant or useful. In this particular case, there is a big red warning on the EPE page stating that there is a discrepancy between the expected and observed formulae, viz. "Expected formula : C8 H18 N2 O4 S1 ... Observed formula : C6 N2 O3 S1" - the discrepancy (ignoring hydrogens) being "C2 O1" which corresponds rather well with the missing hydroxyethyl group.
If one follows the link to ChemPDB, one should be able to get complete coordinates. Other links on the HIC-Up pages that provide links to other PDB entries with the same compound are the list at the top of each compound page, the RCSB list, the PDBSUM list, and the EDS statistics page (which also gives you access to the electron density). If you find a better copy of your hetero compound in another structure, get the coordinates and throw them into the PRODRG server and, optionally, input the PRODRG-regularised coordinates to the HIC-Up server to get, e.g., an O dictionary.
In X-PLOR/CNS, the electrostatic interactions have not been used by default for crystallographic refinement for many years because they can introduce horrible artefacts (e.g., rigid-body refinement can fail merely because two arginines get in each other's way). Hence, the charge estimates are irrelevant for crystallographic applications. Hopefully, anyone who does want to include charges would notice that all charges have been set to zero. PRODRG produces an estimate of the charges so one can run a compound through that server instead (e.g., with the ready-made "Run PRODRG" button on every compound's HIC-Up page).
The server version of XPLO2D (and the one that generates all dictionaries in HIC-Up) strips all hydrogen atoms it encounters (unless they are incognito). This may not go down well with our NMR brethren. There is a (somewhat roundabout) solution, though:
A genuinely frequently-asked question on many bulletin boards is of the type: "I would like to know whether someone has topology and parameter files for acarbose for the CNS suite. If so, could you please E-mail them to me?".
Try the following recipe (a.k.a. "DVD's E-Z Guide to Top and Par Files for Heteros"):
A step-by-step description for CNS-newbies (and not-so-newbiew) of how to include a (covalently-linked) hetero-compound in CNS refinements is provided by Alan Purvis (here).
Another description is provided by Bryan Lepore (here; previously hosted at Brandeis, but now here at the HIC-Up site - with thanks to Bryan!).
Another frequently-asked question is of the type: "I have a problem. I have two sulfates in my structure and when I refine the protein including the sulfates using the minimize option of CNS the sulfates lose the tetrahedric conformation and three of the four oxygens adopt a planar disposition. I use the parameter and topology files from HIC-Up. Someone knows which the problem is?".
From HIC-Up (or XPLO2D or the HIC-Up server) you always get a little energy minimisation input file. Run just that file through CNS (since the resulting conformation will be the best you can hope for once you include the macromolecule and the experimental data !). If you get a distorted compound, your dictionary is to blame, and you'll need to fix it before you use it in refinement. If the geometry looks fine, there is some other problem.
The following question was asked on the O-info mailing list:
"I want to generate the appropriate O & CNS dictionary files for a novel ligand analogue. It is not in HIC-Up. How do I begin ?"
"First you'll need a reliable set of coordinates, e.g. from the CSD, some quantum-chemical program, or a 2D->3D converter. Then you can run it through the dictionary generator either at HIC-Up (link) or at PRODRG (link). PRODRG can also do both steps at once as long as you don't have too exoticelements in your compound."
Carsten Schubert provided the following, more useful answer:
"Ahhh, looks like we have a serious case of the "Industrial Crystallographer Blues" here. The short answer is use Gerard's XPLO2D.
The long answer is: Draw the molecule in a modeling program and minimize it. Alternatively run a conformational search procedure and select a couple of representative structures. Write it/them out as a PDB file, remove unnecessary junk from the PDB, like connect and hydrogens. In case of multiple models give each conformer its own residue number and concatenate their PDB-files, no BREAK or TER cards. Feed the final PDB through XPLO2D, option AUTO. Use either X or _ as an appendage to the chemical atom type. If you want an O dictionary this option only works with Ov7, I believe. Alwyn is still in the process of integrating a couple of things into the stereochemistry dictionary in Ov8 or later. XPLO2D as not been adapted to this and Gerard is waiting for Alwyn to come to closure (hint...). Anyhow at this stage it would really help if you know something about CNS topology and parameter files. XPLO2D writes out a .top and .par file and a couple of PDB files. Check if all your atom types and presumed bond types (single, double ...) are correct, edit and rerun the _xplo2d file if needed. Run the provided CNS input script and check if your minimized geometry works out (flat aromatics, correct stereo centers ...) If not, XPLO2D is sometimes overzealous in defining dihedrals. Get rid of duplicate definitions, and make sure you assign each atom its on chemical atom type in XPLO2D. If you have unusual atoms in your ligand like some pesky chlorines or fluorines, you have to edit the scattering library in CNS, so it picks up the right chemical atom type, because neither FX or CLX are defined."