Distance geometry based comparative modelling

A distance geometry based protein modelling algorithm is presented which relies on the projection of simple model chain coordinates into Euclidean spaces with gradually decreasing dimensionality. Fast embedding was achieved by performing separate distance matrix projections on subsets of the model points. Structural equivalences between the unknown target and related proteins with known structures were deduced either from a mixed sequence/structure multiple alignment or from the output of various fold recognition (threading) approaches. These equivalences were mapped onto the model as structure-specific conserved C alpha atom distances and secondary structure assignments. Additional nonspecific distance restraints derived from general stereochemical properties of folded protein chains were used to guide the modelling process. The method quickly constructed a large number of low-resolution models which could then serve as starting conformations for full-atom refinement. Structure predictions for some targets in the 'Asilomar Challenge' (CASP2) are presented to illustrate potential applications of the approach.