We consider an extension of classical data analysis into symbolic data analysis to describe the management process of biological datasets produced by multi-source clinical studies. Such extension leads to more complex data types and tables and the metadata under consideration hold information both on classical (original) and the symbolic data. In this paper we model these metadata items in a structured object-oriented schema for symbolic data revealing their relations. A number of transformations are also discussed both for classical and symbolic classes of our model in order to illustrate how the applied transformations on symbolic data depend on the related classical data setting.