MIfold

MIfold predicts RNA secondary structures from multiple sequence alignments by using mutual information to identify covarying sites and to support detection of simple pseudoknots.

Key Features:

• Mutual Information Utilization: Uses mutual information and related covariation measures to analyze multiple sequence alignments and identify covarying nucleotide positions.
• Dynamic Programming Algorithm: Employs a dynamic programming algorithm that predicts secondary structures by maximizing the total mutual information across an alignment.
• Pseudoknot Prediction Capability: Predicts simple pseudoknots within RNA secondary structures.
• Sensitivity and Selectivity Adjustment: Provides adjustable performance settings to prioritize sensitivity or selectivity for predictions.
• Performance Enhancement with Sequence Number: Prediction effectiveness improves as the number of aligned sequences increases, enhancing covariation signal.

Scientific Applications:

• Complementary analysis: Offers complementary sensitivity characteristics relative to programs such as RNA Structure Logo, RNAalifold, and COVE.
• Database annotation: Generates automatic structural predictions useful for annotation of RNA families in databases such as Rfam.

Methodology:

Applies mutual information and related covariation measures to multiple sequence alignments and uses a dynamic programming algorithm to predict secondary structures that maximize total mutual information; includes routines for predicting simple pseudoknots and adjusting sensitivity versus selectivity.