RNAG

RNAG predicts consensus secondary structures for unaligned RNA sequences using a blocked Gibbs sampling algorithm to infer structural ensembles and characterize prediction uncertainty.

Key Features:

• Blocked Gibbs Sampling Algorithm: Iteratively samples from P(Structure | Alignment) and P(Alignment | Structure) for faster convergence.
• Posterior Space Characterization: Quantifies prediction uncertainty by analyzing variability in sampled structures.
• Enhanced Prediction Accuracy: Demonstrates higher accuracy on three public datasets with compact sampled structures around ensemble centroids.
• Cluster Analysis: Identifies distinct structural clusters in 11 of 17 RNA families, revealing functional diversity.
• Quantitative Assessment: Measures structural deviation relative to ensemble variation, capturing significant features missed by traditional methods.

Scientific Applications:

• RNA Function and Evolution: Analyzes consensus structures in RNA families for functional insights and motif discovery.

Methodology:

Uses blocked Gibbs sampling that iteratively samples P(Structure | Alignment) and P(Alignment | Structure) from unaligned RNA sequences and analyzes variability in sampled structures to characterize the posterior space.