Representing data as networks have been shown to be a powerful approach for data analysis in biodiversity, e.g., interactions among organisms; relations among genes and phenotypes, etc. In this context, databases and repositories following a graph model (e.g., RDF) have been increasingly used to interconnect information and to support network-driven analysis. Usually, this kind of analysis requires gathering together and linking data from several distinct and heterogeneous sources. In this work, we investigate this challenge in the context of biological bases focusing on the characterization of living organisms, especially their phenotypes and diseases. It includes the rich diversity of Model Organism Databases (MODs) { repositories specialized in a particular taxon { widely used in the biological and medical studies. We exploit a lightweight integration approach, inspired in the Linked Open Data initiative, mapping several biological bases in a uni ed graph database { our BioGraph { and linking key elements to o er an interconnected view over the data.
2. We have created a uni ed database containing data from all these data sources. To solve the heterogeneity problem, we developed a uni ed model to support di erent approaches to describe phenotypes. 3. We have interlinked data from several sources combining two strategies: (i) exploiting existing cross references among sources; (ii) importing bridges between ontologies: Uberon and Uberpheno. 4. With the interlinked graph, we have inferred new edges and nodes, generating knowledge.
The main contributions of this work are: the uni ed model to support several descriptive approaches for phenotypes and the uni ed graph database, containing descriptions of phenotypes from 63 distinct data sources. Future work includes: to import genes, linking them with their phenotypes and diseases and to implement an interface for our system.