Background: Molecular biology (MB) is a dynamic research domain that benefits greatly from the use of modern software technology in preparing experiments, analyzing acquired data, and even performing "in-silico" analyses. As ever new findings change the face of this domain, software for MB has to be sufficiently flexible to accommodate these changes. At the same time, however, the efficient development of high-quality and interoperable software requires a stable model of concepts for the subject domain and their relations. The result of these two contradictory requirements is increased complexity in the development of MB software.A common means to reduce complexity is to consider only a small part of the domain, instead of the domain as a whole. As a result, small, specialized programs develop their own domain understanding. They often use one of the numerous data formats or implement proprietary data models. This makes it difficult to incorporate the results of different programs, which is needed by many users in order to work with the software efficiently. The data conversions required to achieve interoperability involve more than just type conversion. Usually they also require complex data mappings and lead to a loss of information.
Results: To address these problems, we have developed a flexible computer model for the MB domain that supports both changeability and interoperability. This model describes concepts of MB in a formal manner and provides a comprehensive view on it. In this model, we adapted the design pattern "Dynamic Object Model" by using meta data and association classes.A small, highly abstract class model, named "operational model," defines the scope of the software system. An object model, named "knowledge model," describes concrete concepts of the MB domain. The structure of the knowledge model is described by a meta model. We proved our model to be stable, flexible, and useful by implementing a prototype of an MB software framework based on the proposed model.
Conclusion: Stability and flexibility of the domain model is achieved by its separation into two model parts, the operational model and the knowledge model. These parts are connected by the meta model of the knowledge model to the whole domain model. This approach makes it possible to comply with the requirements of interoperability and flexibility in MB.