Development of statistical and computational methods, tools, and infrastructure as well as data analysis, data management, and support for clinical researchers
Final Report Abstract
The aim of the interdisciplinary research group (CRU) was to understand why patients with rectal cancer respond differently to standard therapy in order to advance the development of a therapy tailored to the individual patient. Specifically, aims for the CRU were the following: 1. Manage the data of the CRU and provide databases and infrastructure 2. Provide biostatistical and bioinformatics data analysis support and develop and implement the required methodology As part of the CRU the first aim of this subproject was to establish a professional IT-Infrastructure according to GCP guidelines and German privacy law. Two databases were implemented: one for clinical data and one for biomaterial data. In the course of the two funding periods, five different projects were implemented in the clinical database (one validation study). The implementation of these databases enabled the researchers to perform different data queries for different data analyzes but also for data quality assurance. In order to assess the options to reuse health care data in this clinical research setting, the implementation of a research-focused form in the Göttingen clinical workspace (electronic patient record) was piloted, and a customized form to report quality specific events was set-up. Unfortunately, it turned out that the currently used IT System for the clinical workspace did support the form, but did not offer a sufficient reporting tool to actually use the data. A new approach will be followed based on the KFO experience as soon as an improved electronic health record will be in place at the UMG. The second aim of this subproject was to provide biostatistics and bioinformatics support for the entire KFO. Clinical data and data from laboratory research were collected and analyzed with appropriate statistical and bioinformatics methodology. Major aims were to test for biomarkers that can predict outcome of CRT or disease progression. Different types of molecular data were utilized here. Low dimensional data from immunohistochemistry as well as basic clinical data from the time of biopsy or the time of surgery were used to predict progression-free survival. High-dimensional data from gene expression microarrays or methylation were used to predict outcome after CRT. Further analysis of more targeted experiments in cell lines were used to analyze the basic mechanisms and pathways involved in CRT-resistance. The collaborative efforts led to many discoveries of relevant biomarkers and disease mechanisms in rectal cancer, which resulted in a multitude of peer-reviewed publications and several clinical trials. Besides the aim to provide support in data analysis for the KFO, another aim of this subproject was to establish the required biostatistics and bioinformatics methodology on campus and to test and improve methodology. The methodological research focused on three topics: 1. Analytic methods for microarray data; 2. Methods of risk prediction for cancer patients; 3. Reconstruction and analysis of signaling pathways. In all of these three aims, we could develop and publish relevant methods that enhance the field of bioinformatics and make it more available in medical applications. We develop classification methods for building disease signatures from multiple data. We implemented reconstruction methods for pathways and regulation mechanisms involved in colorectal cancer. We developed and enhanced classification procedures that allow multiple or ordinal responses.
Publications
- Dynamic Deterministic Effect Propagation Networks: learning signalling pathways from longitudinal protein array data. Bioinformatics. 2010;26(18):i596-i602
Bender C, Henjes F, Fröhlich H, Wiemann S, Korf U, Beißbarth T
(See online at https://doi.org/10.1093/bioinformatics/btq385) - Integration Of Pathway Knowledge Into A Reweighted Recursive Feature Elimination Approach For Risk Stratification Of Cancer Patients. Bioinformatics. 2010;26(17):2136-44
Johannes M, Brase JC, Fröhlich H, Gade S, Gehrman M, Fälth M, Sültmann H, Beißbarth T
(See online at https://doi.org/10.1093/bioinformatics/btq345) - Comparison of Global Tests for Functional Gene Sets in Two-Group Designs and Selection of Potentially Effect-causing Genes. Bioinformatics. 2011;27(10):1377-1383
Jung K, Becker B, Brunner B, Beißbarth T
(See online at https://doi.org/10.1093/bioinformatics/btr152) - pathClass: An R-Package for Integration of Pathway Knowledge into Support Vector Machines for Biomarker Discovery. Bioinformatics. 2011;27(10):1442-1443
Johannes M, Fröhlich H, Sültmann H, Beißbarth T
(See online at https://doi.org/10.1093/bioinformatics/btr157) - Adaption of the Global Test Idea to Proteomics Data with Missing Values. Bioinformatics. 2014;30(10):1424-30
Jung K, Dihazi H, Bibi A, Dihazi GH, Beißbarth T
(See online at https://doi.org/10.1093/bioinformatics/btu062)
