Abstract
This thesis studies from an algorithmic point of view various consensus methods on collections of labeled objects. The problems under study involve labeled objects without repetition of labels ; these objects may be rooted trees or sequences, with applications to bioinformatics. For instance, the problems on trees considered in this thesis may find applications to the estimation of congruence between phylogenies, the construction of supertrees, and the identification of horizontal gene transfers. For their part, the problems on sequences considered in this thesis have potential applications for the computation of genomic distances based on gene orders. Overall, this work relies on the theories of parameterized complexity and approximability to obtain algorithms and hardness results for the problems studied.