Large scale molecular biology experiments, the study of biological pathway networks, the analysis of genome sequences, and more in general the analysis and modelling of complex biological systems require advanced data management technologies to access and integrate massive amounts of data coming from different, heterogeneous sources. Recently, semantic technologies have been successfully adopted for this purpose. In particular, Knowledge Graphs are being used as a flexible mechanism that provides a uniform representation and access to heterogeneous information. Such graphs are complemented by an ontology, which encodes complex domain knowledge that helps enriching the answers obtained when querying the actual data. We consider the Virtual Knowledge Graph (VKG) paradigm, where the graph is exposed virtually by means of declarative mappings that specify how the concepts represented in the ontology relate to data sources. In the talk, we describe the principles underlying the VKG approach to data management, and illustrate on some meaningful examples from the biomedical domain how it can be applied to effectively access and integrate data.