Abstract
A new algorithm is described that forms a single structure representative of ensembles of structures from files in the format used by the Protein Data Bank. A first attempt is made by averaging in the space spanned by bond lengths, inter-bond rotations and symmetry-multiplied dihedral rotations. This normally produces well formed regular secondary-structure elements, but the intervening less well ordered regions are often distorted because of the invalidity of averaging large rotations about divergent axes. For this reason, the algorithm includes a second stage that pulls the interatomic distances towards more fully representative values. Results produced by this method have proved better as judged by conventional quality checks than any input structure in nearly all cases tested so far, especially for the backbone, and much better than those produced by commonly used alternative methods.
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call