Improved measurements of RNA structure conservation with generalized centroid estimators

Area of Science:

• Bioinformatics
• Molecular Cell Biology
• Genomics

Background:

• Non-protein-coding RNAs (ncRNAs) are vital in molecular cell biology and bioinformatics.
• ncRNA function is intrinsically linked to secondary structure, making its identification crucial.
• Existing methods like minimum free energy (MFE) are biased by nucleotide composition, and measures like Structure Conservation Index (SCI) and Base Pair Distance (BPD) can yield high false discovery rates, particularly in genome-wide searches.

Purpose of the Study:

• To develop improved metrics for ncRNA identification based on SCI and BPD.
• To enhance robustness against low-quality multiple sequence alignments using generalized centroid estimators.
• To provide more accurate and reliable ncRNA identification, especially for large-scale genomic analyses.

Main Methods:

• Application of generalized centroid estimators to SCI and BPD.
• Incorporation of robustness against low-quality multiple sequence alignments.
• Experimental validation using both human-curated and computationally generated alignments (CLUSTAL W, RAF).

Main Results:

• Proposed methods demonstrate higher accuracy than original SCI and BPD on both curated and low-quality alignments.
• Centroid-based SCI on CLUSTAL W alignments shows comparable or superior performance to original SCI on high-quality RAF alignments.
• The new methods are computationally efficient, requiring less time than high-quality aligners.

Conclusions:

• The developed centroid-based SCI and BPD offer superior accuracy and robustness for ncRNA identification.
• These improved methods are particularly well-suited for analyzing low-quality, genome-wide alignments.
• The findings advance ncRNA discovery in bioinformatics and molecular cell biology.