Provides an automated system for the continuous evaluation of RNA structure prediction methods. The goal of CompaRNA is to provide a ‘blind’ benchmark using experimental data before it becomes incorporated in the training data set of the assessed prediction methods. In analogy to protein structure prediction benchmarks Livebench and EVA, CompaRNA uses as a reference experimentally solved RNA structures deposited in the Protein Data Bank (PDB). It also provides the community with a ‘static’ benchmark, generated on a data set extracted from the RNAstrand database, which contains experimentally verified RNA secondary structures of any type and organism. RNAstrand includes a larger number of molecules than those with solved 3D structures, and it covers sequences that are on the average longer than those with known 3D structures. These benchmarks offer insight into the relative performance of different RNA secondary structure prediction methods on different types of RNA sequences and structures.
This work was supported by the European Research Council (ERC, StG grant RNA+P=123D), the Polish Ministry of Science and Higher Education (currently National Science Center (NCN), grant N N301 035539), European Social Fund (‘Scholarship support for PhD students specializing in majors strategic for Wielkopolska’s development’, Sub-measure 8.2.2 Human Capital Operational Programme, co-financed by European Union under the European Social Fund), the Deutscher Akademischer Austauschdienst (D/09/42768), and an ‘Ideas for Poland’ fellowship from the Foundation for Polish Science.