Artificial intelligence (AI)-based algorithms and technologies are spreading throughout scientific research, driving a paradigm shift from hypothesis- to data-driven science. In the biomedical field, which deals with complex, high-dimensional biological systems, the enormous increase in information availability has introduced computational challenges as well as new possibilities for knowledge discovery. In this respect, graph-based models are powerful tools to represent multilevel complex structures and semantic information. A classification system of a large set of organisms based on metabolic networks discrimination will be illustrated as a sound approach in the emerging field of microbiome analysis. A comprehensive method will be shown, for reconstructing the 3D whole-genome chromatin structural ensemble from Hi-C data, paving the way to studies on chromatin dynamics and function/regulation-related conformational alterations.

In a reverse perspective, AI architectures have been successfully inspired by biological processes and structures: it will be shown the example of “analog” neuromorphic computing which, complementing the accuracy of conventional computing, may outperform it in cognitive tasks such as planning, decision making and sensory processing.
Notwithstanding the advantages of the beneficial synergy between biology and AI, issues still remain before routine use of AI systems in the research and clinical field: it will be presented the successful case of a recently constituted multidisciplinary team directly active on site.