Résumé
Sequencing technologies have dramatically evolved during the last years, making available the whole genome sequence of an increasing number of species. This evolution has open a new field of researches, and a scale change in our data analyses, with new bioinformatic tools developped to manage large sets of data. This context has change our view to elaborate cocoa breeding strategy. Two cocoa genome sequences have been produced recently: the first one published and completely annotated was established from a Criollo genotype originated from Belize ; 28798 genes coding for protein were identified ; the other sequence was established from a Forastero genotype originated from Brazil. Many other cocoa sequences, corresponding to expressed genes were also produced, and a wide field of new applications is now offered for cocoa breeding. The first immediate application is the production of a huge number of new SSR and SNP genetic markers easily identified along the genome sequence, and genotyped using high throughput technologies. Sequence comparison between the 2 genomes and/or expressed sequence tags allow to generate thousands of molecular markers which could be located very precisely in the cocoa genome. One of the last impressing method is the GBS (genotyping by sequencing technology), which allow to generate simultaneously several thousands of SNP markers and genotype each plant for a reasonable price.These dense marker distribution along all the genome, now accessible, change our view to carry out cocoa selection and will allow to better integrate markers in a global breeding approach such as a recurent breeding strategy. Marker assisted strategies, aiming to accumulate favorable traits with the help of markers, have evolved in the last years with these new technologies, taking advantage of a possible use of a very high density of markers widespread along the genome. A new strategy, called Genomic Selection (GS), relies on an integrative and whole genome marker assisted selection approach which takes in account all marker information along the genome. The GS is particularly adapted for the selection of complex and polygenic traits depending to many genes with small effects. With the availability of large sets of markers, localised along the genome, a genomic selection is now possible for cocoa and can be applied to improve populations, as those managed during cocoa recurent breeding steps. Such a strategy is presently tested to increase yield and disease resistance level of cocoa genotypes in Brazil and Cameroun and to improve quality traits in Ecuador. Another main application is the help of the genome sequence to identify candidate genes involved in cocoa useful trait variation to elaborate " diagnostic tools " for specific traits: a link can be first established between the QTL and the corresponding genome sequence region, and fine mapping approaches is then developped to precise the identification of candidate genes responsable of the trait variation. A direct PCR amplification from cocoa organs, now possible, could then be used to easily establish the diagnostic. Synteny among species of the vegetal kingdom is now well established, and particularly at the level of gene structure and function. Comparative genomics with model plants, made now easier with the availability of the genome sequence, is another key approach to progress more rapidly in the knowledge of cocoa gene function and in cocoa breeding. (Résumé d'auteur)