The pennate diatom Phaeodactylum tricornutum is a model organism able to synthesise industrially-relevant molecules. Large-scale monocultures are prone to bio-contamination, however, little is known about the identity of the invading organisms. To gain insight into the bacterial community associated with diatoms, we translated the complexity of a natural system into reproducible experiments where we investigated the microbiome of P. tricornutum cultures. The results revealed a dynamic bacterial community that changed over time and in differing media conditions. We propose a network of putative interactions between P. tricornutum and the main bacterial factions, which is translated into a set of ordinary differential equations constituting a computational dynamic model. The proposed mathematical model is able to capture the population dynamics, further supporting the hypothesised interactions. The interdisciplinary approach implemented provides a framework for understanding the dynamics of diatom-associated microbial communities, and provides a foundation for further systematic investigations of host-microbe interactions.