Semantic Web intro + COEUS

DOI: 10.1145/2064747.2064768

Presentation

Abstract
Relevant biomedical advances happen daily, and the medical profession relies on this evolution to deliver an improved patient care. In addition, the growing magnitude of data generated by biomedical software and hardware since the initial discovery of the human genome is remarkable in size and variety. Hence, best-of-breed software solutions are at best a couple years behind clinical practice demands.

In this paper we detail an innovative Semantic Web interoperability framework, which provides developers with a complete software stack for semantic application deployment. Interoperability is the defining feature of this framework. On the one hand, new instances are able to integrate several types of distributed and heterogeneous data. On the other hand, collected data are made available through a public SPARQL endpoint.

Semantic Web developments are gaining momentum in the scientific research community: the usage of semantics is finally being considered as a valuable solution to integrate resources and compose services.

The Semantic Web can be seen as a set of new ideas regarding the way we think about our data and, subsequently, about our applications. Nowadays, we are used to model and plan our systems around the data we have and the features we want to offer. Semantic Web forces a shift in these thoughts: we should be thinking about the meaning of our data combined with the meaning of its relations with other systems and what we can extract from these combinations. Moreover, traditional strategies can hardly cope with the dimension of our current problems. The amount of data we want to integrate and the type of features we wish to offer are exactly the ones that motivate Semantic Web developments.

How does it work?

To express meaning we need to define statements. Statements are simple Subject-Predicate-Object triples. For instance: Alice lives with Bob or P05067 is a Protein. Some authors go as far as saying that using these triples we can express everything we want: we just need to find the right combination between Subject, Predicate and Object. By combining sets of triples, we enable knowledge, which is exactly what we want to gather!

In the previous figure we have various statements. First, we say that uniprot:P05067 is a Protein. Next we say that it is named Amyloid Precursor Protein. On the right side we have omim:104300 which is Disease called Alzheimer. Next by combining uniprot:P05067 with omim:104300 and stating that uniprot:P05067 is involved in omim:104300 we are enabling new knowledge: the Amyloid Precursor Protein is involved in the Alzeimer disease. By expanding this strategy to billions of concepts and relations we will be able to extract deeper and more insightful information from our data sources.

From a technology point of view, Semantic Web can be very similar to current systems. We have an ontology (or various) which defines the meaning of our data and how it is related: the data model. We need data storage, using RDF, to store our ontology, our data and our relations: the database. At last, we need to able to query our data storage to answer our questions. To do this we have SPARQL: a SQL-like query language.

Why GEN2PHEN?

In my opinion, GEN2PHEN is the perfect scenario for novel Semantic Web developments. The goals behind this project are directly tied with extracting and discovering knowledge (we have a Knowledge Centre) and with the integration of multiple resources. And this is where Semantic Web strategies may play a key role: they were designed for it!Semantic Web developments in GEN2PHEN can be organized in two levels.

First, the study of existing and the creation of new ontologies. This can help in the standardized definition of various fields, related to genotype and phenotype research. These ontologies could be used within the GEN2PHEN project or by other developers in the life sciences community.

Second, the Knowledge Centre could provide a semantic database and a set of services, allowing access and reasoning over the integrated semantic data, thus, allowing the development of next-generation bioinformatics applications.

Goals

The goals behind this interest group will be achieved in two stages. For now, I would like this group to be used for ideas and experiences exchange.

Is anyone using Semantic Web? What are you doing? How are you doing it? What problems have you faced? Do you have any doubts and want to start developing? I would really like to know if some partner is doing any research in this field and if it can relate with GEN2PHEN.

In a long term perspective, I envisage the discussion of more concrete topics (like in the Web Services group for instance) which would lead to the created of the previously mentioned GEN2PHEN semantic database.

Above all, this group should raise awareness of this “brave new world” that is the Semantic Web inside GEN2PHEN and, who knows, empower the development of some new applications which can be helpful to the life sciences community.

Final application published at DETI Journal 2010.

Keynote for the Information Visualization course of my doctoral programme

Keynote for the Bioinformatics course of my doctoral programme