The focus of our research at the University of Zurich is the analysis of structural variations in cancer genomes by computational genomics, including bioinformatics and systems biology methods. Our work centers around our collections of molecular tumor data, assembled from genomic screening experiments in cancer e.g. through Comparative Cytogenetic Hybridization (CGH) studies. Specific projects deal with the development of computational methods for structural data analysis, genomic aberration analysis in tumor entities as well as with the large scale exploration of genomic patterns in malignancies.
The neoplastic transformation of cells is based on the accumulation of genomic mutations disrupting the tight regulation of apoptosis, growth, proliferation and microenvironmental interaction. So far, mostly the effects of single gene mutations have been defined. Not much is known about the implications of extensive genomic aberrations introduced e.g. through genomic copy number abnormalities, which can be observed in most cancer entities. With the growing recognition of similar mutations in clinically unrelated entities (e.g. BRAF mutations in astrocytomas and melanomas, HER2 activation in breast but also colorectal cancer), research targeting only a single tumor type clearly cannot address some of the intriguing aspects related shared oncogenetic mechanisms. Our group works on the curation of large scale oncogenomic data sets, and on developing the necessary mining, analysis and visualization tools for cross-platform/cross-entity oncogenomic meta-analyses.
Projects and Services
Quality assessment and data integration of multi-platform genomic array data
- Data processing pipelines for oncogenomic arrays from various platforms, including large insert clone arrays, oligonucleotide and SNP arrays: The focus here is not on low-level processing methods but on building robust multi-platform solutions.
- Automated rejection methods for low quality array data: Data from public repositories frequently is “contaminated” through low quality samples, for which so far no cross-platform evaluation methods exist.
- Calibration of copy number data and CN level assignment: While gain/loss status data can usually be derived based on empirical thresholds, the assignment of absolute regional allele counts presents an unsolved problem.
- Probe (re-) mapping matrices: The majority of deposited genomic array data does not come with explicit probe position assignments according to a current genome edition. Having performed extensive map assignments during data analysis, we intend to share this information in an upcoming repository, with updates for future genome editions.
Functional and integrative analysis of oncogenomic data
- Statistical correlation in CNA data: We analyse and weigh e.g. CNA co-occurrences, and project relevant data points onto functional interaction networks.
- Cancer type specific CNA detection: We develop and apply statistical methods to identify copy number events with high disease specificity.
- Pathway enrichment analyses from CNA data: We test methods to identify pathways whose genes are statistically enriched as targets of copy number changes in cancer.
- Multi-dimensional data integration: We explore other gene specific data qualities and there integration into our cancer CNA data analysis strategies.
Targeting cancer: The genetics of childhood brain tumors
- Medulloblastomas and PNET tumors: Immature neuroectodermal tumors are, after acute leukemias, the second most frequent class of childhood malignancies. With the recent recognition of the genetic/biological heterogeneity of these entities, we aim to provide reference information for a molecular characterization and stable classification of these diseases.
- Childhood gliomas and DIPG: Childhood gliomas are a rare group of tumors which differ from the adult counterparts. Among those, DIPG (diffuse intrinsic pontine gliomas) comprise a specific subtype with clinically dismal prognosis. Especially due to the rare occurrence of these cancers, the curation of all observed instances and meta-analysis of data will be essential for understanding the tumors' biology.
Data repositories & online tools
Progenetix - genomic copy number aberrations in cancer
The Progenetix database provides an overview of genomic copy number abnormalities (CNA) in human cancer. As a curated database, it collects CNA and associated clinical data of individual cancer and leukemia cases, from peer reviewed studies.
As of July 2012, Progenetix contains 20400 chromosomal CGH and 9383 genomic array experiments from 363 distinct diagnostic entities. It is the largest public database for interpreted ("called") whole genome CNA profiles. While the website provides entity and series specific CNA frequency profiles, online tools allow for user directed data selection and visualization.
arrayMap - genomic arrays in cancer
arrayMap is a curated reference database and bioinformatics resource, targeting copy number profiling data in human cancer. The database provides an entry point for meta-analysis and systems level data integration of high-resolution oncogenomic data. The current data reflects more than 40000 oncogenomic array experiments, from approximately 200 cancer entities.
In contrast to Progenetix, arrayMap presents experiment specific data rather than called/interpreted, tumor specific results. Probe level visualization as well as customized data representation facilitate gene level and genome wide data review. Results from multi-case selections can be connected to downstream data analysis and visualization tools.
DIPGdata - genomic data repository for Diffuse Intrinsic Pontine Gliomas
The DIPG Genomics Repository is an International collaboration supported by The Cure Starts Now Foundation. It aims to provide a central resource for researchers to investigate genome-wide profiling data from childhood diffuse intrinsic pontine glioma specimens. This work forms part of a systematic review and meta-analysis of paediatric glioma genomics, aimed at collating publicly-available data sets of these diseases in children.
Websites for Further InformationComputational Oncogenomics group: www.imls.uzh.ch/research/baudis.html