This work sets out a methodological approach to assess the relationships that exist between four sets of data: breeding practices, carcass characteristics, m. Longissimus thoracis biomarkers and m. Semitendinosus biomarkers. Seventy-one young bulls of 3 breeds were characterized by 78 variables. The variables in each set were arranged into 4, 3, 5 and 3 homogeneous clusters respectively using the clustOfVar approach. A second clustering of variables established from these 15 clusters was then used to build 5 pools of clusters that were precisely described to characterize the interaction between the 4 sets of variables.Some practices have a universal impact on biomarker patterns eg. increasing straw proportion in the diet leads to an increase in MyLC-1F and Hsp40 biomarkers whatever the breed and the muscle. Nevertheless, most breeding practices appear to have varying impacts on biomarkers depending on animal type and muscle. For example, ST biomarkers are less related to breeding practices and/or carcass characteristics than in LT. Moreover, some biomarker expressions appear to be breed-dependant, especially for the Angus breed which is very specific in comparison to continental breeds. The breed where muscle biomarkers are the most closely linked to breeding practices and carcass characteristics is the Blonde d’Aquitaine. In this breed, MDH1 and DJ-1 biomarkers are positively linked to muscle yield and carcass development in both muscles and some other LT biomarkers might also be considered as good indicators of carcass fatness (small Hsp, ENO3, Capz-β, MyBP-H).< Réduire