AlignmentFold and AlignmentPartition: Improving the align-then-fold approach for RNA secondary structure prediction

Area of Science:

• Computational Biology
• Bioinformatics
• Molecular Biology

Background:

• Predicting conserved RNA secondary structure from homologous sequences is crucial.
• Existing methods often combine free energy minimization with sequence covariation analysis.
• Fixed alignments are commonly used, but introduce challenges like gaps and non-canonical base pairs.

Purpose of the Study:

• To develop novel computational tools, AlignmentFold and AlignmentPartition, for enhanced RNA secondary structure prediction.
• To refine thermodynamic models by including parameters for gaps and non-canonical base pairs.
• To evaluate the impact of alignment quality and covariation analysis on prediction accuracy.

Main Methods:

• Development of AlignmentFold for consensus minimum free energy structure prediction.
• Development of AlignmentPartition for base pairing probability prediction.
• Determination of new nearest neighbor thermodynamic parameters for gaps and non-canonical base pairs.
• Assessment of prediction accuracy based on alignment methods, size, and inclusion of covariation analysis.

Main Results:

• AlignmentFold and AlignmentPartition predict RNA secondary structures and base pairing probabilities.
• New thermodynamic parameters were determined for gaps and non-canonical base pairs.
• Prediction accuracy was evaluated against various alignment strategies and covariation analyses.
• Excluding covariation from structure quality scoring did not reduce, and potentially improved, prediction accuracy compared to other tools.

Conclusions:

• AlignmentFold and AlignmentPartition offer accurate RNA secondary structure prediction.
• The developed thermodynamic parameters enhance predictions for alignments with gaps and non-canonical pairs.
• These tools are available within the RNAstructure software package, facilitating broader research applications.