Hundreds of databases offer vast amounts of literature knowledge about biological signaling networks. However, this knowledge is rarely integrated into current bioinformatic analyzes due to challenges in the programmatic access and transformation of this data. This thesis focuses on the integration of prior knowledge into methods for network reconstruction. The motivation is to improve the performance of bioinformatic algorithms and methods by facilitating the integration of available pathway data as prior knowledge. First, the fundamentals of biological networks and pathways, their encoding using ontologies, methods for network reconstruction, and high-throughput gene expression technologies are introduced. Three central results are presented in this work: First, the novel software package rBiopaxParser, which enables the generic import of BioPAX-encoded pathway databases into the R Project for Statistical Computing. An overview of the functionality, the internal data model and visulization options is given. Second, a proof-of-concept implementation of the transformation and merging of pathway data to be used as prior knowledge for methods for network reconstruction is presented. The interactomes, the entirety of interactions, of three databases, Reactome, Pathway Interaction Database, and BioCarta, are generated and merged as a basis for prior pathway knowledge. Third, network reconstruction using Nested Effects Models is performed based on the generated prior knowledge networks and experimental high-throughput data of 16 gene knockdowns in breast cancer cell lines. Finally, this thesis compares the implemented software to similar concurrent developments and discusses the generated prior knowledge and the results of network reconstruction.
