The Dead-End Elimination Theorem: Mathematical Aspects, Implementation, Optimizations, Evaluation and Performance

The placement of amino acid side chains in a given fixed main-chain template forms a recurrent but nontrivial task in protein modeling. Even for a small set of side chains in a given protein, the degrees of freedom for the side chains lead to an enormous number of combinatorial possibilities, inevitably prohibiting a brute force approach to pinpoint the global minimum energy conformation (GMEC). The recognition of the existence of statistically relevant discrete combinations of the dihedral angles (called rotamers) of a side chain forms the basis of all current side-chain placement techniques (1). Several research groups have published methods to predict the side chain positions in a fixed protein main-chain trace (review in refs. 2 and 3)