A knowledge-driven approach for crystallographic protein model completion.

ABSTRACT: One of the most cumbersome and time-demanding tasks in completing a protein model is building short missing regions or ;loops'. A method is presented that uses structural and electron-density information to build the most likely conformations of such loops. Using the distribution of angles and dihedral angles in pentapeptides as the driving parameters, a set of possible conformations for the C(alpha) backbone of loops was generated. The most likely candidate is then selected in a hierarchical manner: new and stronger restraints are added while the loop is built. The weight of the electron-density correlation relative to geometrical considerations is gradually increased until the most likely loop is selected on map correlation alone. To conclude, the loop is refined against the electron den

SUBMITTER: Joosten K 

PROVIDER: S-EPMC2467521 | biostudies-literature | 2008 Apr

REPOSITORIES: biostudies-literature