About the project

PredPharmTox is an e-infrastructure for Large Scale Predictive Pharmacology and Toxicology, providing easy access to chemical and omics data in toxicology and pharmacology, together with methods and predictions available via user-friendly tools. The infrastructure will support open standards and be developed in close collaboration with European initiatives, and implemented as a Platform as a Service (PaaS) linking to the cloud resource of the Swedish National Infrastructure for Computing (SNIC). The infrastructure will enable Swedish researchers to perform world class science such as integrated pharmacology/toxicology approaches, chemical safety assessment, toxicogenomics and pharmacogenomics with mechanistic interpretations and discovery of chemically-initiated pathways, as well as including omics data in predictive toxicology and pharmacology modeling.



PredPharmTox is based on a VR-infrastructure planning project from 2011 (dnr: 2011-6129). The project started in January 2012 and will continue until the end of 2013. Continued funding has been sought for 2014-2017 from VR.

Prototype infrastructure

We have established an instance of the framework Galaxy as an embryo for the PredPharmTox e-infrastructrue, available from http://galaxy.predpharmtox.org. To demonstrate the power of such an e-infrastructure we have made a simplistic integration of (parts of) the datasets TG-GATES for toxicogenomics, STITCH for chemical-protein interactions, and STRING for protein-protein interactions. We have also developed two methods for querying the data sources, allowing for relating local data (in form of sets of genes or chemicals) to the integrated datasets.

Application 1: ToxMap

ToxMap provides a visual representation of the toxicogenomic space occupied by a query signature, using state-of-the-art statistical tools for matching to over 2600 TG-GATEs human in vitro dose-response profiles. Users are able to submit a gene signature (or a gene expression profile) and obtains results in form of a graphical visualization as well as similarity results in tabular formats.

P. Kohonen, S. Lampa, R. Ceder, R. Grafström, and O. Spjuth. Toxicity Map: From Gene-Expression Signatures to a Prediction of a Dose Response to a Chemical. Submitted.


Application 2: QuantMap

QuantMap QuantMap is a method for assessing relative bioactivity of chemical substances using quantitative molecular network topology analysis, and which originally was developed by co-authors MG and UH. We reimplemented the method in order to be automated and used in batch analysis, as well as externalized the data sources to allow for future extensions. QuantMap takes a list of compounds as input and delivers a dendrogram based on bioactivity similarity.

W. Schaal, M. Gustafsson, U. Hammerling, and O. Spjuth. Automated QuantMap for rapid quantitative molecular network topology analysis. Bioinformatics 2013, 29(18):2369-70. doi: 10.1093/bioinformatics/btt390



Architecture of the PredPharmTox e-infrastructure. Central to the infrastructure are the semantic databases (grey color) which can be accessed directly via a SPARQL client but more user-friendly via the informatics platform implemented on top of the Galaxy framework as a Platform as a Service on a private cloud, with a connection to SNIC cloud resource for scaling up computational and storage resources on demand. Red boxes denote ways to interact with the e-infrastructure and include a Web API for programmatic access, a Web browser for accessing the Galaxy web portal, and the Bioclipse workbench which is a rich client (downloadable application).