Refinement of proteins (e.g. to get standard uncertainties on distances)

  • go to http://shelx.uni-ac.gwdg.de/SHELX and read "SHELX-97 Manual as PDF", "Mini-protein refinement tutorial" as well as "P1-Lysozyme refinement tutorial", "Thomas Schneider's FAQs" and "FAQs: Macromolecules"
  • run shelxpro to obtain .ins file from .pdb file; a ligand may require the "J" option
  • use "CGLS x y" refinement until convergence; the last run should be "CGLS x" only.
  • a final job to get standard uncertainties (s.u., formerly e.s.d.) on all geometric parameters (see Q21 in "FAQs: Macromolecules"):
    • change CGLS x y to REM CGLS x y
    • insert lines L.S. 1, DAMP 0 0 and BLOC 1 (or e.g. BLOC N_1 > LAST )
    • remove all restraints: lines begining with SIMU, DELU, ISOR, BUMP, DFIX, DANG, CHIV, FLAT (from "Mini-protein refinement tutorial").
      (Editor's) Note a): NCSY then should also be removed, right? b) all of this is only useful for high-resolution work (let's say 1.4 A). At low to medium resolution, my typical question would rather be: what are the s.u. on the unrestrained distances between atoms x and y, given that these atoms are covalently bound to other atoms with restraints? In other words, I would keep all geometric restraints and ask "how much is the s.u. on this unrestrained distance based on crystallographic data (reflections to some resolution) for this model (with its R-factor, B-factors and restraints)?".
    • BOND, RTAB, HTAB and MPLA instructions may be needed to define the dependent parameters for which esds are required (from "FAQs: Macromolecules"). As an example, BIND FE_5001 NE2_123 together with BOND FE_5001 NE2_123 would enter the distance between FE_5001 and NE2_123 into the connectivity table, and would print out the distance and its s.u. into the .lst file.