Abstract

A combination of graphics and computational chemistry gives insight into the refinement process and subsequent analysis of macromolecular structure. The techniques involved have been applied to proteins under investigation in this lab. Ribbon drawings represent and elegant means to visualize the global features of macromolecules. Color-coding of ribbons by mainchain dihedral angles, sensibility of sidechain chi values (from crystallographic database analysis) and degree of fit to electron density maps provide useful clues for refinement. The space curve that underlies ribbon construction has interesting properties. Data base analysis of amino acid sidechain surface preference in a set of refined protein structures has allowed the identification of low probability residue contacts. The CEDAR energy minimization and molecular dynamics program allows viewing of a structure color coded by bond or angle strain, or by van der Walls or electrostatic forces. The programs can also map out the potential fields of active site's volumes. These potential fields are displayed as contour maps or converted into ridge-line networks for a minimalist representation that allows dynamic contouring. Graphics display is done primarily with the programs FRODO and GRINCH, both funning on VAX/VMS/PS300 vector and SGI/UNIX/IRIS raster units. Computations are generally done on the IRIS. The PSFRODO firmware is easily modified to allow inclusion of extra graphics objects, such as multiple ridge-line networks or molecular dynamics animation.

Ray-traced ORTEP drawing.
A nucleoside analog solved in the lab.

Ridge-line energy mapping.
Maps the trypsin/benzamidine interaction as a ridge line network. The size of the spheres gives their relative energy. Green is hydrophobic interactions, blue where H-bond donors want to be, and red for H-bond acceptors.

Sidechain preference factor analysis.
This compares the surface-based approach (which includes water) to the distance-based approach of S.Bryant. Only 2 factors are significant in explaining the side chain preferences: 50% explained by factor-1 plotted along the X-axis, and <20% for factor-2 plotted along Y. See if you can figure out what these factors are. (See the database chapter for actual preference data).

Model of Apamin.
A model of apamin was proposed by the late Alan Zell in this lab, based on a hunch about the disulphide bonding patterns observed in scorpion toxins. We never could get this published, but his model was essentially confirmed by later NMR studies.