Data integration and generation of a phenotype-protein-drug network

Coordinator:    Dr. Miguel Andrade
Institution: Max-Delbrück-Centrum für Molekulare Medizin (MDC) Berlin-Buch
The goal of this subproject was the development of project databases for all experimental data generated by the project partners. We developed methods to combine this information with external data available in the public domain to predict novel associations between genes, drugs and neural diseases.On the basis of phenotype-genotype relations, we predicted new associations between (i) genes and diseases, (ii) protein-protein interactions (PPIs) or functional modules and diseases, and (iii) drugs that have effects on genes or functional modules.The NeuroNet database has been developed to make the data generated in the project accessible and searchable (manuscript in preparation). The HIPPIE database was developed to integrate our results with experimentally verified PPI data (Schaefer et al., PLoS One, 2012; Schaefer et al., PLoS Comp Biol, 2013; von Eichborn et al., Nucleic Acids Res, 2013). This allowed us to obtain insight into the mechanisms of protein aggregation in neural diseases (Petrakis et al., BioEssays, 2013; Petrakis et al., PLoS Genetics, 2012; Schaefer et al., Nucleic Acids Res, 2012). We developed data and text mining methods to integrate experimental and disease information derived from the bibliography with gene information (Ortuño et al., BMC Bioinformatics, 2013; Fontaine et al., Nucleic Acids Res, 2011; Barbosa-Silva et al., BMC Bioinformatics, 2011; Donard et al., BMC Genomics, 2011). In collaboration with our clinical partners (Charité Berlin; manuscript in preparation) we are exploring such methods to define genotype-phenotype relations in neural diseases to produce novel predictions of drug effects on disease. We generated novel methods for the analysis of Y2H data, which we applied to neural-disease related proteins (Suter et al., Nucleic Acids Res., 2013), and we analysed RNAseq data comparing the blood of Huntington’s patients to uncover blood markers for Huntington’s disease.This project was facilitated by close co-operation and exchange of data and ideas with the team of Prof. Dr. Peer Bork (EMBL  Heidelberg).

Figure 1:  Proposed mechanism for polyglutamine (polyQ) function. Left, a protein has a polyQ region C-terminally from a coiled-coil region. Right, upon an interaction between coiled-coils the polyQ region modulates the interaction by extending the coiled coil. We achieved evidence for this model after studying the network of PPIs in human and model organisms and observing correlations between polyQ regions, predicted coiled-coil regions, and number of interacting partners for proteins with such features.

Further Coordinators:
INTRANET (Members login)