An efficient two-level swarm intelligence approach for RNA secondary structure prediction with bi-objective minimum free energy scores

This paper introduces a set-basedtwo-level particle swarm optimization algorithm (TL-PSOfold) with multiple swarmsfor finding secondary structure of RNA with prediction accuracy. First objectiveis concerned with maximizing number of stacked loops at hydrogen bond, whereas, second objective deals with minimum free energy (MFE) at standard nearest neighbor database (NNDB). First level of the algorithm works on theen tire search space for the best solution of each swarm, where as, these cond level worksatthe gbest solution of each swarm.The set based PSO approach has been applied at both levels to represent and up date these to for dered pairs of the folded RNA sequence. Improved weight parameters chemes with mutation operators are implemented for better convergence and to over come the stagnation problem. Bi-objectives nature of TL-PSO fold enables the algorithm to achieve maximum matching pairs as well as optimum structure a trespective levels. The performance of TL-PSO fold is compared with a family of PSO based a lgorithmsi. e. Helix PSO v1, Helix PSO v2, PSO fold, Set PSO,IPSO, FPSO, popular secondary structure prediction software RNA fold, mfold and other metaheuristics RNA-Predict, SARNA-Predict at the criteria of sensitivity, specificity and F-measure. Simulation results for TL-PSO fold show that it yield shigher prediction accuracy than all the compared approaches. The claimis supported by the non-parametric statistical significance testing using Kruskala-Wallistest followed by post-hoc analysis.