Electron density maps o When looking at PDB structures Electrons Density (ED) maps are more/as important as the 3D atomic model! o ED is a 3D map of where the scattering electron cloud is according to the measured X-ray data. o 2Fo-Fc map indicates where electrons are (according to SF and model). Normally colour blue or grey. o Fo-Fc difference map: green for positive difference: where the current model fails to place sufficient electrons Red for negative difference: where the current model places too many electrons www.ebi.ac.uk/pdbe/entry/pdb/4z9l/bound/ANP Electron density for a ligand with poor fit 2h7p.pdb: ED visualized in coot Notice difference density around ligand 2h7p has been obsoleted and replaced 4tzt with really superb ED and ligand fit 2h7p.pdb: ED around ligand, as visualized in buster-report http://grade.globalphasing.org/tut/erice_workshop/ Data resolution affects electron density detail: Well placed/refined sucrose ligand at different data resolutions: 3 1ylt 1.2Å resolution “atomic resolution” 2Fo-Fc ED shows atoms as Individual blobs. Need higher resolution for hydrogen atoms 2pwe 2.0Å resolution Typical medium resolution for ligand studies. Can see ring pucker 2qqv 3.0Å resolution Low resolution. Ligand placement unambiguous but fine detail cannot be seen Model improvement • Basically interpret electron density maps in real space to improve model • Initially automated methods (warp/arpwarp) used • But mostly manual corrections to the model are done using the Coot program • Look at Fo-Fc difference map for both negative and positive features • Build into 2Fo-Fc Why do crystallographers make mistakes? • Limitations to the data • Incomplete • Weak • Limited resolution • Space and time averaged • The human factor • Subjectivity and bias involved in map interpretation and refinement (even at atomic resolution!) • Inexperienced people do the work, use of black boxes, … • Not everybody is a good chemist • Even experienced people make mistakes Kleywegt, Acta Cryst. D65, 134 (2009) PDF report for depositor & referees Statistics and plots for the entry, per chain, per residue, and list of unusual features wwPDB X-ray validation pipeline Validation pipeline 1.0 MolProbity EDSXtriage Mogul Deposited data (coordinates & reflections) Percentiles PDF maker Validation XML file Distributions External reference files (e.g., Engh & Huber) Gore et al., Acta Cryst. D68, 478 (2012) wwPDB validation reports • Pipeline produces PDF report and XML output • Slider graphic useful • Current PDF is “rather verbose” pdbe.org/valrep/1cbs Ramachandran plot • Look at main chain dihedral angles phi and psi • Ramachandran et al. (1963) worked out only certain combinations of phi/psi cause clashes https://en.wikipedia.org/wiki/Ramachandran_plot Rotation around φ with ψ=0o Rotation around ψ with φ=0o (Images kindly provided by David Sanders, University of Saskatchewan.) Summary ‘Sliders’ Validation information for users Atoms bumping into each other Surprising torsion angles Atoms not in electron density Sidechain outliers • Just like the main chain phi and psi dihedral angles amino acid sides chains have chi angles with have preferred and disallowed regions The 5 chi angles of an arginine side chain http://www.ccp14.ac.uk/ccp/web-mirrors/garlic/garlic-1.5/commands/dihedrals.html Summary ‘Sliders’ Validation information for users Atoms bumping into each other Surprising torsion angle Atoms not in electron density MolProbity – clash score • Idea is to look for bad nonbonded contacts after hydrogen atoms have been added to the model • Very powerful method • Suggests NQH flips • Included in wwPDB validation reports • Or Use from: • Molprobity web site • Or within coot http://molprobity.biochem.duke.edu/ Validation information for users Summary ‘Sliders’ Atoms bumping into each other Surprising torsion angles Atoms not in electron density Real-space fit • Quantitative, real-space measure of how well a residue fits its local density (Jones et al., 1991) • Express as R-value (RSR) or correlation coefficient (RSCC) • RSR = S | robs - rcalc | / S | robs + rcalc | • Sums extend over all grid points inside a mask around the residue RSR - real-space R-value pdbe.org/1cbs • RSR = S |robs - rcalc| / S |robs + rcalc| Example: retinoic acid (REA) in 1CBS (1.8Å) RSR - real-space R-value Define mask (1.1Å) ~500 Grid points Observed density Calculated density RSR = S |robs - rcalc| / S |robs + rcalc| pdbe.org/1cbs RSRZ is reported in Summary ‘Sliders’ Atoms bumping into each other Surprising torsion angles Atoms not in electron density PDBe simplification of validation sliders “Best molecule” – integration of validation information in PDBe query system www.ebi.ac.uk/pdbe/entry/search NMR validation • NMR VTF recommendations published • Global quality scores reported for“welldefined residues” only • As averages over the ensemble • Medoid model only Montelione et al., Structure 21, 1563 (2013) EM validation reports • Prototype EM map-validation reports • Most of the PDBe “Visual analysis” functionality implemented Compare to: pdbe.org/emd-8116/analysis Ligands in proteins • So you have successfully navigated all the hazards so far have great data, well integrated, successful MR, refinement model building, Ramachandran analysis …… • You have density in the active site and the whole point of the structure is to find how the interesting drug candidate ligand binds ……. • Here be dragons!