Background: Macromolecular structures are modeled by conformational optimization within experimental and knowledge-based restraints. Discrete restraint-based sampling generates high-quality structures within these restraints and facilitates further refinement in a continuous all-atom energy landscape. This approach has been used successfully for protein loop modeling, comparative modeling and electron density fitting in X-ray crystallography.
Results: Here we present a software toolkit (Rappertk) which generalizes discrete restraint-based sampling for use in structural biology. Modular design and multi-layered architecture enables Rappertk to sample conformations of any macromolecule at many levels of detail and within a variety of experimental restraints. Performance against a Calpha-tracing benchmark shows that the efficiency has not suffered despite the overhead required by this flexibility. We demonstrate the toolkit's capabilities by building high-quality beta-sheets and by introducing restraint-driven sampling. RNA sampling is demonstrated by rebuilding a protein-RNA interface. Ability to construct arbitrary ligands is used in sampling protein-ligand interfaces within electron density. Finally, secondary structure and shape information derived from EM are combined to generate multiple conformations of a protein consistent with the observed density.
Conclusion: Through its modular design and ease of use, Rappertk enables exploration of a wide variety of interesting avenues in structural biology. This toolkit, with illustrative examples, is freely available to academic users from http://www-cryst.bioc.cam.ac.uk/~swanand/mysite/rtk/index.html.