This chapter presents the computational prediction of the secondary structures within the 5' and 3' untranslated regions of the dengue virus serotype 2 (DENV2), with the focus on the conformational prediction of the two dumbbell-like structures, 5' DB and 3' DB, found in the core region of the 3' untranslated region of DENV2. For secondary structure prediction purposes we used a 719 nt-long subgenomic RNA construct from DENV2, which we refer to as the minigenome. The construct combines the 5'-most 226 nt from the 5' UTR and a fragment of the capsid coding region with the last 42 nt from the non-structural protein NS5 coding region and the 451 nt of the 3' UTR. This minigenome has been shown to contain the elements needed for translation, as well as negative strand RNA synthesis. We present the Massively Parallel Genetic Algorithm MPGAfold, a non-deterministic algorithm, that was used to predict the secondary structures of the DENV2 719 nt long minigenome construct, as well as our computational workbench called StructureLab that was used to interactively explore the solution spaces produced by MPGAfold. The MPGAfold algorithm is first introduced at the conceptual level. Then specific parameters guiding its performance are discussed and illustrated with a representative selection of the results from the study. Plots of the solution spaces generated by MPGAfold illustrate the algorithm, while selected secondary structures focus on variable formation of the dumbbell structures and other identified structural motifs. They also serve as illustrations of some of the capabilities of the StructureLab workbench. Results of the computational structure determination calculations are discussed and compared to the experimental data.