CONUS

CONUS predicts RNA secondary structures using simple stochastic context-free grammars (SCFGs) and integrates probabilistic evolutionary and biophysical information to balance model complexity and prediction accuracy.

Key Features:

• Stochastic Context-Free Grammars (SCFGs): Uses simple stochastic context-free grammars (SCFGs) to model RNA secondary structure with probabilistic rules.
• Integration of Evolutionary and Biophysical Data: Integrates evolutionary and biophysical probabilistic information to inform and constrain structure prediction.
• Model Simplicity and Efficiency: Emphasizes small, simple SCFG designs to limit the number of free parameters while maintaining predictive accuracy.
• Exploration of SCFG Designs: Implements nine distinct SCFG designs to evaluate trade-offs between model simplicity and prediction performance.
• Benchmark Testing: Tests each SCFG design against a benchmark set of RNA secondary structures to assess single-sequence prediction accuracy.
• Comparable Accuracy to Energy Minimization: Identifies four SCFG designs that achieve prediction accuracies comparable to current energy minimization programs.
• Knudsen and Hein PFOLD Design: Includes the Knudsen and Hein design used in the PFOLD algorithm, noted for using 21 free parameters.

Scientific Applications:

• Comparative RNA Sequence Analysis: Supports comparative RNA sequence analysis by incorporating evolutionary models into structure prediction.
• Biophysical Modeling: Informs biophysical modeling of RNA by evaluating the plausibility of predicted secondary structures using biophysical data.
• Single-Sequence Structure Prediction: Applicable to single-sequence RNA secondary structure prediction and performance benchmarking.

Methodology:

Implements multiple SCFG designs with a focus on minimal free parameters, combines probabilistic evolutionary and biophysical data to inform predictions, and conducts benchmark testing of single-sequence structure prediction performance.