These applications are meant for researchers interested in evaluating structure prediction protocols with respect to local structural features or researchers interested in improving a structural biology energy function.

We provide tutorials that describe an example of a concrete task as a way to orient new users. At the end of each tutorial there references where more detailed information is provided. If there is a specific tutorial you think or would like to see here, please add it to the list at the bottom of requested tutorials!

We also document individual FeatureReporters and R scripts useful for post-processing data.

Overview

In order to aid in analysis and comparisons of native protein structures and Rosetta models and designs, the Rosetta Feature Reporter framework was developed comprising an analytical framework developed in Rosetta C++ and a comparison and plotting framework developed in R, a free, open-source comprehensive statistical package. The framework was initially used to improve the default Rosetta energy function and hydrogen bonding (resulting in the now-default talaris2013 energy function), but has since been expanded for use in a number of analytical and comparative tasks. In the framework, a Feature is a component or set of components of a structure to be analyzed. These ‘features’ encompass energies, hydrogen bond distances, sequences, etc.

A FeaturesReporter is responsible for reading and writing a feature or set of features to a relational database (mySQL, SQLITE3, etc.). RosettaScripts is used to specify which FeatureReporters to run on a set of structures.

Comparing a set of features from two different sets of structures is done in two steps. The first step is to analyze each set of structures using a chosen number of FeatureReporters through an XML script and the RosettaScripts application. Each FeatureReporter is its own Rosetta C++ class and analyzes a set of features from the input structure and outputs one or more tables into a relational database.

Once two or more databases are created for a set of features derived from each structure in each dataset, the second step is to compare the datasets using the FeatureReporter R Framework. This framework enables one to specify the input databases to compare as well as various output options and the set of Feature R Scripts to be run through either the command line or a JSON file specifying the components (a common data interchange format used by a variety of computational languages and programs). Each Features R script requires the use of different FeatureReporters to enable the comparison, with some R scripts requiring many FeatureReporters to be used. The plots and tables output by the FeatureReporter R Framework can then be used to deduce similarities and differences between the given datasets in addition to various statistical outputs.

A features database contains all of the structural information associated with a set structures in a batch of structures. The hierarchy below outlines the general organization of this database.

              protocols      <------------ One row per execution of Rosetta
                 /\
                 ||
         /---> batches <---\    <--------- One row per set of structures
        / |      /\       | \                
       /  |      ||       |  \
      /   |      ||       |   \
     /-----> Structures <------\   <------ One row per structure, uuid 
    /     |    /     \    |     \
 Feature  Feature   Feature   Feature  <-- Each FeatureReporter manages a type of    
 Reporter Reporter  Reporter  Reporter     of structural data. Static data is indexed 
                                           per batch.

References

O'Meara, M. J., Leaver-Fay, A., Tyka, M., Stein, A., Houlihan, K., DiMaio, F., Bradley, P., Kortemme, T., Baker, D., Snoeyink, J., A Combined Covalent-Electrostatic Model of Hydrogen Bonding Improves Structure Prediction with Rosetta . Journal of Chemical Theory and Computation, 2015.

Leaver-Fay, A., O'Meara, M. J., Tyka, M., Jacak, R., Song, Y., Kellogg, E. H., Thompson, J., Davis, I. W., Pache, R. A., Lyskov, S., Gray, J. J., Kortemme, T., Richardson, J. S., Havranek, J. J., Snoeyink, J., Baker, D., Kuhlman, B., Scientific benchmarks for guiding macromolecular energy function improvement. Methods in enzymology, 2013. 523: p. 109.

See Also