Conclusions

Looking Back

The field of computational protein design has progressed rapidly in the last few decades. Like all new technologies, early efforts focused on achieving noticeable results with any system that could be used reliably. Designing metal-binding sites was an attractive target since such sites can be described in simple geometric terms, and near exhaustive searches could be performed within a reasonable period considering the processing power of computers at the time. Since many transition metals have spectroscopic signatures when chelated by particular ligands in specific geometries, testing the designs was straightforward. As the field progressed, geometry-based designs began to shift focus toward creating proteins with technological applications, most notably in the development of biosensors.

Advances in algorithms for searching sequence space and improved force fields to describe atomic interactions have made stereochemistry-based design of complex surfaces possible. Again, early efforts focused on tractable problems— having a strong assay for protein activity or binding often dictated which protein was used as a template. With new reports of progressively more intricate designs being published every year, computational design is starting to be seen as a realistic alternative or complement to diversity-oriented approaches. Furthermore, the software for designing proteins is becoming more accessible to the general scientific community. Many of the software packages used in the 1990s and early 2000s were (and still are) command line/script-based programs that require expert training and extensive programming knowledge to set the design parameters correctly.

Looking Forward

The field of computational design is changing. Some of the programs are available for download or are accessible online. However, the design process is, for the most part, still not an interactive experience. This is likely to change in the near future. PyMol (http://www.pymol.org) is an open-source structure visualization program that runs on all main platforms and has a very intuitive interface. In addition to providing built-in tools for visualizing mutagenesis, plug-ins for various applications are available. As software becomes more user-friendly, computational protein design is likely to become more accessible to biochemists with less computational experience. With the RosettaDesign server (http://rosettadesign.med.unc.edu) (Liu and Kuhlman 2006) being accessible via a PyMol plug-in, the future could see protein engineers visualize a structure, pick the residues they want to mutate with a mouse, choose optimization criteria, and have a design returned to them shortly thereafter. Reliable "point-and-click" protein design will surely reduce the time it takes to design/ improve/modify proteins, but the real value of computational design is in its ability to surpass the physical and technical limitations of diversity-oriented design. When those designs improve the quality of life of people, success will be real.

0 0

Post a comment