The field of Generative Models has achieved impressive results in recent years. This project aims to keep investigating this field, focusing primarily on methodological and theoretical aspects of diffusion models in a first phase. In particular, we will explore how to incorporate ideas from Informed Machine Learning, which proposes to leverage prior knowledge and empirical data to enhance model performance. Eventually, applications of these informed generative models to the biology and health domains will be explored, as they often involve complex and intricate systems that could benefit from such techniques.