Knowledge management for systems biology and translational medicine. Experiences from the EU BioBridge project

Abstract

The integrative approach of Systems Biology promises a better understanding of complex phenotypes such as cardiovascular disease, diabetes or chronic obstructive pulmonary disease (COPD). It aims to interconnect current knowledge with experimental data, in-silico analysis and simulation. To this end the existing knowledge must be semantically integrated in a single environment and dynamically organised into structured networks that are connected with experimental data. From the resulting information local sub-networks with associated experimental data need to be extracted. The EU BioBridge project coordinates clinical, experimental and computational groups generating a Systems Biology workflow for the analysis and therapeutic intervention of COPD. In order to address the integration, extraction and reporting challenge we deployed the BioXM knowledge management environment as central infrastructure for the BioBridge project. BioXM allows the dynamic, graphic generation of domain specific knowledge representation models based on specific objects and their relations supporting annotations and ontologies. Wizards automatically adapted to the created knowledge model allow to map data from external sources for use as federated resource or for import. The Java based client-server application thus provides a flexible and customisable framework for knowledge management. BioXM implements visual browse and query interfaces used to traverse the knowledge network and construct complex queries and retrievals. Using BioXM a COPD specific instance of the system has been set-up as part of the BioBridge project http://www.biobridge.eu/bio/. It allows users to mine COPD specific molecular networks, clinical and experimental data and provides pre-structured queries and reports to retrieve sub-networks spanning protein-protein interaction, pathway, gene - disease and gene -compound data for subsequent data analysis, model building and simulation. In the current version, the knowledge base integrates more than 20 different databases and ontologies representing a total of 80 793 genes (30 246 human, 27 237 mouse, 23 310 rat), 1 307 pathways, 78 528 compounds, 1 525 474 protein interactions and the entire Gene Expression Omnibus database resulting in a total of 3 666 313 connections within the knowledge network. This public information is used for the integrative analysis of project specific clinical e.g. questionnaires, anthropometric and physiologic data with experimental e.g. gene expression and metabolimcs data and subject specific literature derived molecular knowledge. A command line and a SOAP web service API allow to integrate BioXM into larger bioinformatic infrastructures. Analytical applications such as R-scripts can be integrated transparently as native BioXM views or analyses. Within BioBridge we created an SBML based integration with modelling and simulation tools such as MathModellica, IsoDyn and ByoDyn which allow to generate and simulate deterministic models. To parameterise and constrain the models, these are combined with data analysis tools such as BANJO and ARACHNE which allow the generation of probabilistic networks from expression, metabolomic and proteomic data. We will present technical information about the BioBrige infrastructure as well as describing the COPD specific resources which become publicly available to researches.