In this thesis, we have presented a novel approach to interoperability based on the use of biological relationships that have used relationship-based integration to integrate bioinformatics data sources; this refers to the use of different relationship types with different relationship closeness values to link gene expression datasets with other information available in public bioinformatics data sources. These relationships provide flexible linkage for biologists to discover linked data across the biological universe. Relationship closeness is a variable used to measure the closeness of the biological entities in a relationship and is a characteristic of the relationship. The novelty of this approach is that it allows a user to link a gene expression dataset with heterogeneous data sources dynamically and flexibly to facilitate comparative genomics investigations. Our research has demonstrated that using different relationships allows biologists to analyze experimental datasets in different ways, shorten the time needed to analyze the datasets and provide an easier way to undertake this analysis. Thus, it provides more power to biologists to do experimentations using changing threshold values and linkage types. This is achieved in our framework by introducing the Soft Link Model (SLM) and a Relationship Knowledge Base (RKB), which is built and used by SLM. Integration and Data Mining Bioinformatics Data sources system (IDMBD) is implemented as a proof of concept prototype to demonstrate the technique of linkages described in the thesis.