Comparative Study of Single-stranded Oligonucleotides Secondary Structure Prediction Tools.

Area of Science:

• Bioinformatics
• Computational Biology
• Molecular Biology

Background:

• Single-stranded nucleic acids (ssNAs) play vital biological roles and have significant biotechnological potential due to their diverse spatial conformations.
• The secondary structure, or base pairing pattern, is the primary determinant of ssNA spatial organization.
• Numerous computational tools exist for predicting ssNA secondary structures, creating a need for a comparative guide on their efficacy and limitations.

Purpose of the Study:

• To evaluate and compare the performance of nine freely available computational tools for predicting ssNA secondary structures.
• To identify the most accurate and reliable tools for secondary structure prediction based on experimental data.

Main Methods:

• A comparative analysis of nine ssNA secondary structure prediction tools: mfold, RNAfold, CentroidFold, CONTRAfold, MC-Fold, LinearFold, UFold, SPOT-RNA, and MXfold2.
• Performance evaluation was conducted on a dataset comprising 538 ssNAs with experimentally determined secondary structures.

Main Results:

• Minimum free energy (MFE) based tools (mfold, RNAfold) and artificial intelligence (AI) tools (CONTRAfold, MXfold2) demonstrated the highest accuracy in exact predictions.
• MC-fold exhibited the lowest performance.
• UFold and SPOT-RNA were the only tools capable of predicting pseudoknots.
• Considering suboptimal solutions improved the accuracy of mfold and RNAfold.
• Challenges remain in predicting specific motifs like multi-way junctions and mini-dumbbells, and structures in complex with proteins.

Conclusions:

• Mfold, RNAfold, and MXfold2 are currently recommended for general ssNA secondary structure prediction, despite limitations with specific motifs.
• AI-driven tools like UFold and SPOT-RNA show significant potential for overcoming prediction challenges, particularly for pseudoknots, and represent a promising future direction.