Computational protein design: a novel path to future protein drugs

Computational protein design emerges in recent years as a field that could make a substantial impact on the design of protein drugs. It still consists mainly of redesigning parts of a protein sequence for increasing the stability of a given 3-dimensional conformation of a protein, but has already been extended from redesigning core residues to redesigning in all other protein regions, as well as to the design of backbone conformations. More recently, proteins with new binding functions and new enzymes, protein libraries, designs of full folds and of a new protein fold, have been some of the main highlights. The search and the scoring problems are however not fully solved, and many of the design processes should be examined on much larger scales in order to assess their usefulness. We examine some of the basic assumptions in computational protein design, in particular, the separation between sequence and scaffold designs. Among others, we suggest to include more protein residues in computations, to include relevant parts of the backbone, to use appropriate reference states, to produce the proteins and to validate the designs by structural examination of the protein products.