PARTS: probabilistic alignment for RNA joinT secondary structure prediction

Area of Science:

• Computational Biology
• Bioinformatics
• Molecular Biology

Background:

• Predicting RNA secondary structure and sequence alignment is crucial for understanding RNA function.
• Existing methods often use restrictive structural alignment models, limiting their ability to account for RNA structural variability.

Purpose of the Study:

• To develop a novel method for joint prediction of RNA sequence alignment and common secondary structures.
• To introduce a generalized structural alignment framework that accommodates greater structural variability within RNA families.

Main Methods:

• Introduced 'matched helical regions' for a generalized structural alignment search space.
• Utilized a probabilistic model with pseudo free energies for scoring alignments.
• Employed a dynamic programming algorithm (PARTS) for Maximum a posteriori (MAP) predictions.

Main Results:

• PARTS demonstrated significantly improved secondary structure and alignment predictions for the RNase P family compared to prior methods.
• Performance was comparable to existing methods for tRNA and 5S rRNA families.
• PARTS improved posterior probability estimates and achieved superior sensitivity and positive predictive value for base pairings over single-sequence predictions.

Conclusions:

• The PARTS method offers a more general structural alignment approach, enhancing joint RNA sequence and structure prediction accuracy for certain RNA families.
• PARTS provides improved confidence estimates for base pairings, outperforming single-sequence prediction methods.