A knowledge federation strategy for integrating genotype-to-phenotype data

The life sciences research field is witnessing an explosive growth in data generated by next-generation genotyping technologies. With this increase, it will be exponentially harder to detect meaningful sequence variations in our genome. Understanding these data, our genotype and the connections to rare disease proneness or expected drug response, our phenotype, is essential to ensure future personalized medicine scenarios. Ongoing research projects adopt a “divide-and-conquer” approach, being geared towards the study of particular diseases or specific genome loci. For this matter, ad-hoc software solutions are developed, focusing on a limited set of features often targeted at niche fields. Consequently, the resulting software ecosystem is fragmented and heterogeneous.
The meaningful data connections innate to Semantic Web solutions mimic the biological domain relationships amongst distinct entities. Semantic Web technologies’ maturity promoted true meaningful and autonomous connections amongst data, and brought with it new strategies to tackle life sciences challenges. In spite of the increased awareness regarding Semantic Web technologies’ advantages, its adoption has been slow paced in the biological domain. With a steep learning curve and lacking turnkey solutions, developers are faced with many roadblocks, resulting in a low number of purely semantic solutions. Our work introduces a new application deployment framework, providing bioinformatics with a complete software stack for creating customizable Semantic Web applications. Each new instance will include a native federation layer, opening the door for harnessing more insightful knowledge from the aggregated network of relationships.

By using this framework, we want to spring new life to genotype-to-phenotype research. With it, current legacy applications will be replaced by modern interoperable systems, and new tools will emerge for the remaining fields, covering the entire genotype-to-phenotype domain.
In an ideal scenario, each application can still cover its own area: genetic mutations, protein interactions, drugs and diseases can each be studied in a distinct system. However, we will be able to obtain a holistic over these distributed datasets, analysing them as a single knowledge base. WAVe, targeted at gene curators and covering the human variome field, and DiseaseCard, a rare disease portal, will be replaced in the initial stage. Later, with multiple instances online, we will be able to connect data located at each independent application, launching a truly federated knowledge network over distributed biological data.

No relevant conflicts of interest declared.