This work package aims to overcome the bioinformatic challenges associated with a potentially transformative technology: high-throughput functional assays on patient-derived samples. As one instance, the systematic mapping of drug response using large, comprehensive panels of drugs and large, representative cohorts of patients’ tumour samples as performed in WP 5 and 6 promises to become a powerful tool for the understanding of disease biology, for the discovery of individual tumour specific vulnerabilities (e.g. specific pathway inhibition, combinations) and for individualized treatment choice [Bock, Nature Rev. Cancer 2012; Tyner, Cancer Res. 2013; Pemovska, Cancer Discov. 2013]. Besides drugs, assays may also employ other genetic perturbation tools including RNAi and in the near future CRISPR-Cas9 [Torres, Nature Comm. 2014]. Already now, such screens comprise tens of thousands of measurements (e.g., hundreds of 384-well plates), but the need to work with large patient cohorts as well as to test for combinations for compounds asks for throughput at even larger scales. Furthermore, there is interest in assays “beyond cell viability”, using more realistic culture systems and more informative readouts such as microscopy, high-content screening, RNA and protein expression signatures.
However, the lack of standard methods to analyse these complex data is a major challenge to the adoption of these approaches by clinical research groups. This WP will deliver the tools to overcome this hurdle. It will develop statistical methods, embedded in an extensible and well-documented bioinformatics package, to analyse high-throughput screens, applicable to a variety of assay types. Specifically, the software will accommodate multi-patient and multi-perturbation designs and multivariate (i.e. “high-content”) phenotypic readouts, and it will be able to link these to multi-omic molecular (genetic, transcriptomics, etc.) characterisation of the samples. Its functionality will include suitable data structures for efficiently and safely managing the data and metadata, visualisation, as well as statistical methodology for proper normalization, dimension reduction, feature selection, statistical testing and modelling, and discovery of interesting patterns.
By providing a robust bioinformatics tool built on modern statistical methodology, the outcome of this WP will contribute greatly to the scale-up, automation and standardization of high-throughput functional assays. This will not only benefit individual projects, but also extend the scientific scope of this type of approach by making it available to a wider range of biomedical researchers, and by empowering better data quality and comparability.
The outcome of this work package will be open-source software that will initially be used by the clinical partner and will be packaged, published and documented such that it can easily be deployed beyond the consortium by the expanding number of researchers interested in functional profiling of patient cells.
- European Molecular Biology Laboratory, Heidelberg (Lead Partner)
- ETH Zurich
- University of Cambridge
- Technische Universität München
- Instituto de Engenharia Mecânica, Lisbon
- German Cancer Research Center, Heidelberg
- Hospital of the University of Zurich