Résumé en anglais

Microarray technology allows for the evaluation of the level of previous termexpressionnext term of thousands of previous termgenesnext term in previous termanext term sample of cells under previous termanext term given condition. In this paper, we introduce previous termanext term methodology based on cooperative previous termGamenext term Theory for the selection of groups of previous termgenesnext term with high power in classifying samples, according to previous termgene expressionnext term patterns. The connection between microarray previous termgames and classification gamesnext term is discussed and the use of the Shapley value to measure the power of previous termgenesnext term for previous termclassificationnext term is motivated on particular instances and compared to the interaction index.