A modular framework to extend Rosetta protocols with multistate design
Patrick Löffler, Samuel Schmitz, Enrico Hupfeld, Rainer Merkl · 2016
Computational protein design (CPD) is a powerful technique to design novel proteins. Many CPD objectives such as design on backbone ensembles, multi-specificity design and the integration of negative design demand the simultaneous optimization of multiple design states. Rosetta [1] is a popular software suite to study and design proteins. Rosetta’s protocols consist of specific procedures and a fine-tuned set of parameters to carry out a given task. An example is the use of specific sequence design cycles and catalytic constraints in the enzyme design protocol. At present, the multistate design implementation of Rosetta [2] is a generic approach lacking options to fine tune the calculations in the same manner as specialized single state protocols. We have developed a framework for CPD that integrates multistate design in existing Rosetta protocols while preserving the protocol’s original functionality. Our framework consists of two, easily exchangeable components: i) The optimizer searches the sequence space and ii) the evaluator scores the sequences according to the given design task. Currently, we utilize Rosetta’s genetic algorithm as an optimizer; protocols for enzyme design [3] and protein-protein interface design [4] serve as evaluators. However, due to the modularity of both components, multistate functionality can be transferred to arbitrary Rosetta applications with little effort. We have benchmarked the enzyme design application on a dataset consisting of conformational ensembles and achieve an 18 percent performance improvement over singlestate methods. As a proof of concept, we have applied our framework to computationally design retro-aldolases which are currently subject to biochemical characterization. References: [1] Leaver-Fay A. et al. (2011). ROSETTA3: an object-oriented software suite for the simulation and design of macromolecules. Methods in Enzymology 487; 545-74 [2] Leaver-Fay A. et al. (2011). A generic program for multistate protein design. PLoS One 6(7), e20937. [3] Richter F. et al. (2011). De novo enzyme design using Rosetta3. PLoS One 6(5), e19230. [4] Lewis S. M. and Kuhlman B. A. (2011). Anchored design of protein-protein interfaces. PLoS One 6(6), e20872.