Malware detection is a crucial aspect of software security. Malware typically recur to a variety of disguise and concealing techniques in order to avoid detection. Metamorphism is the ability of a program to mutate its form yet keeping unchanged its functionality and therefore its danger in case of malware. A major challenge in this field is the development of general automatic/systematic detection techniques that are able to catch the possible variants of a metamorphic malware. We take the position that the key for handling metamorphism relies in a deeper understanding of the semantics of the metamorphic malware. By applying standard formal methods we aim at proving that metamorphic analysis is a special case of program analysis, where the object of computation is code interpreted as a mutational data structure.