Technology to collect and analyse data relating to human health and behaviour will increasingly become a part of everyday life. For the first time, human kind has the opportunity to “time-travel” but not in a mechanical device but through simulation that will allow representations of the past (e.g. to detect causes for disease) as well as glimpse into the future (thus enabling prevention of disease). While the opportunities are mind-blowing, the explosion of data and the need for its analysis also poses many challenges. Clearly, we will need transformative approaches to “make sense” of this data: instead of developing methodology from scratch, we need to look for analogies that will allow us to exploit approaches already existing for other application areas. At the same time, we need to have a keen understanding of what is unique about a given application area, and how existing methodology can be adapted and if needed transformed into a new methodology optimal for providing real-word solutions. This presentation covers theoretical foundations and dealing effectively with the needs arising from specific biomedical applications. We will use specific cases from systems biology to illustrate the data requirements, limitations and possibilities of cutting-edge machine learning approaches. In all cases, integration of data relating to interactions between biological molecules across different organisms and/or diseases is key to understanding the fundamental biological processes underlying diseases and how to treat them. We coin this approach “computational metainteractomics”.