A parallel algorithm for helix mapping between 3D and 1D protein structure using the length constraints

Abstract

Determining 3-dimensional (3D) structures of proteins is still a challenging problem. Certain experimental techniques can produce partial information about protein structures, yet not enough to solve the structure. In this paper, we investigate the problem of relating such partial information to its protein sequence. We developed an algorithm of building a library to map helices in a 3D structure to its 1-dimensional (ID) structure using the length constraints of helices, obtained from such partial information. We present a parallel algorithm for building a mapping tree using dynamic distributed scheduling for load balancing. The algorithm shows near linear speedup for up to 20 processors tested. If the protein secondary structure prediction is good, the library contains a mapping that correctly assigns the majority of the helices in the protein. © Springer-Verlag Berlin Heidelberg 2004.