This thesis is devoted to the design of algorithms that predict RNA secondary structures and related thermodynamic quantities and sequence alignment algorithms. In the first part, a non-Boltmann Monte-Carlo algorithm is used to estimate energy states density or hybridization energy. Our algorithm to estimate density compares favorably to existing algorithms . Computed denaturation temperature are closer to experimental values than values computed by the two pre-existing algorithms. The second part presents an implementation of dynamic programming algorithms that provides sub-optimal structures where riboswitch functional structures are expected. These predictions compare favorably to the prediction of five other algorithms. Third part is devoted to the search of sub-optimal sequence alignments. For proteins, the focus is given to the improvement of alignment quality, given an identity close to 10-15%. In a comparison with Needman-Wunsch algorithm, entropy computed by our program shows a better correlation with reliable residues positions, according to BAliBASE.
