NeuroRDF: Semantic data integration strategies for modeling neurodegenerative diseases

Neurodegenerative diseases are incurable and debilitating conditions with huge social and economical impact, where much is still to be learnt about the underlying molecular mechanisms. Mechanistic disease models could offer a knowledge framework to help decipher the complex interactions that occur at molecular and cellular levels. This motivates the need for development of a framework consisting of heterogeneous data coupled into different regulatory layers. Thus, enabling deeper mechanistic and medical insight into such complex diseases. Here, we describe a methodology to generate semantic web-based mechanistic disease models that allow formalization of complex research questions in order to gain disease understanding. Data for disease model construction was integrated from publicly available (semi-) structured and unstructured data resources into a single semantic web framework called NeuroRDF. Different data types were considered ranging from protein-protein interactions, miRNA-target interactions, pathways, to microarrays. Furthermore, we discuss in detail the data preprocessing effort incurred, and RDF schemas implemented for building NeuroRDF. We illustrate the effectiveness of this approach through a real world biomedical query for biomarker identification in the context of Alzheimer0s disease (AD). Furthermore, we report on the effort and challenges faced during generation of such an indicationspecific knowledge base comprising curated and quality-controlled data.