Structural Modeling of Transcriptomics Data Using Creative Knowledge Discovery

The overall aim of systems biology is to bring a novel perspective into understanding of complex interactions in biological systems. We present a top down approach for modeling of transcriptomics data through information fusion and creative knowledge discovery. By using onotology information as background knowledge for semantic subgroup discovery, rules are constructed that allow recognition of gene groups that are differentially expressed in different types of tissues. This information is further linked with the Biomine engine to visualize gene groups and uncover potential unexpected characteristics of the observed system. In Biomine, data from several publicly available databases were merged into a large graph and a method for link discovery between entities in queries was developed. Obtained models can thus serve as generators of research hypothesis that can be further on experimentally validated. Results of two case studies are presented to illustrate the applicability of the approach.