RNAknot: A new algorithm for RNA secondary structure prediction based on genetic algorithm and GRASP method

Area of Science:

• Bioinformatics
• Computational Biology
• Molecular Biology

Background:

• Predicting RNA secondary structure is a complex computational challenge, especially with diverse pseudoknot classes.
• Existing methods often struggle to balance accuracy and complexity in RNA structure prediction.
• Pseudoknots are crucial RNA topologies involved in various biological processes.

Purpose of the Study:

• To introduce RNAknot, a new computational method for predicting RNA secondary structures.
• To incorporate various structural elements, including stems, loops, and two types of pseudoknots (H-type and Hairpin kissing).
• To enhance the accuracy and efficiency of RNA secondary structure prediction.

Main Methods:

• RNAknot utilizes a combination of a genetic algorithm and the Greedy Randomized Adaptive Search Procedure (GRASP).
• The free energy of RNA structures is employed as the fitness function for evaluation.
• The method was validated using two datasets comprising 26 and 105 RNA sequences from RNAstrand and Pseudobase databases.

Main Results:

• RNAknot demonstrated significantly improved prediction accuracy compared to established programs like Vs_subopt, CyloFold, IPknot, Kinefold, RNAstructure, and Sfold.
• On the first dataset, RNAknot achieved the highest average specificity (71.23%) and sensitivity (72.15%).
• For the second dataset, RNAknot yielded the highest average specificity (85.49%) and F-measure (79.97%).

Conclusions:

• RNAknot offers a superior approach for RNA secondary structure prediction, particularly for sequences with pseudoknots.
• The method's enhanced accuracy and performance make it a valuable tool for bioinformatics research.
• The RNAknot program is publicly available for use in RNA structure analysis.