Riboswitches are RNA genetic control elements that were originally discovered in bacteria and provide a unique mechanism of gene regulation. They work without the participation of proteins and are believed to represent ancient regulatory systems in the evolutionary timescale. One of the biggest challenges in riboswitch research is to find additional eukaryotic riboswitches since more than 20 riboswitch classes have been found in prokaryotes but only one class has been found in eukaryotes. Moreover, this single known class of eukaryotic riboswitch, namely the TPP riboswitch class, has been found in bacteria, archaea, fungi and plants but not in animals. The few examples of eukaryotic riboswitches were identified using sequence-based bioinformatics search methods such as a combination of BLAST and pattern matching techniques that incorporate base-pairing considerations. None of these approaches perform energy minimization structure predictions. There is a clear motivation to develop new bioinformatics methods, aside of the ongoing advances in covariance models, that will sample the sequence search space more flexibly using structural guidance while retaining the computational efficiency of sequence-based methods. We present a new energy minimization approach that transforms structure-based search into a sequence-based search, thereby enabling the utilization of well established sequence-based search utilities such as BLAST and FASTA. The transformation to sequence space is obtained by using an extended inverse RNA folding problem solver with sequence and structure constraints, available within RNAfbinv. Examples in applying the new method are presented for the purine and preQ1 riboswitches. The method is described in detail along with its findings in prokaryotes. Potential uses in finding novel eukaryotic riboswitches and optimizing pre-designed synthetic riboswitches based on ligand simulations are discussed. The method components are freely available for use.