Development of tools to process and analyse precision oncology data in a clinical decision support framework for biomedical research projects in the scope of molecular tumour boards
by Kevin Kornrumpf née Begenat
Date of Examination:2024-10-24
Date of issue:2024-11-12
Advisor:Dr. Jürgen Dönitz
Referee:Prof. Dr. Tim Beißbarth
Referee:PD Dr. Raphael Koch
Persistent Address:
http://dx.doi.org/10.53846/goediss-10854
Files in this item
Name:Kornrumpf_Dissertation_Final.pdf
Size:33.0Mb
Format:PDF
The following license files are associated with this item:
Abstract
English
The goal of precision medicine is to tailor treatments to the individual genetic profile of cancer patients, but it faces several challenges. These include the genetic complexity of tumours, the large volume of data generated by next-generation sequencing (NGS), and the need to integrate multidisciplinary insights from oncologists, bioinformaticians, and other specialists. This work presents major advances in precision medicine, with a focus on the development of computational tools to improve clinical decision-making, particularly in molecular tumour boards (MTBs). By developing innovative tools to process, normalise and interpret genetic variation with a focus on cancer-related genomic data, it closes the gap between bioinformatics and clinical oncology. The first tool discussed in the thesis is SeqCAT, a Sequence Conversion and Analysis Toolbox designed to optimise the process of handling diverse genomic data formats. SeqCAT addresses the complexity of data conversion and ensures consistency and accuracy across datasets, which is critical for subsequent analysis and integration into clinical workflows. This tool enhances the ability of researchers or clinicians working with large genomic datasets to harmonise their data easily without the need of complex mapping rules. Variants of unknown significance (VUS) are genetic alterations whose clinical significance is unclear. They represent a serious challenge to clinical decision-making. The second tool introduced is the VUS-Predict pipeline, which is designed to predict the status of VUS. It utilises machine learning techniques to analyse this particular type of mutation and provides clinicians with probabilistic assessments of whether they are benign or harmful. VUS-Predict supports MTBs in their decision-making and helps to classify the increasing amount of genetic variants. DrugOn is a comprehensive drug ontology designed to support precision oncology. It facilitates the identification of potential alternative treatment options for cancer patients by categorising drugs according to their mechanism of action and therapeutic use. This ontology will help refine the selection of targeted therapies and enable the repurposing of existing drugs for new therapeutic applications, serving as a useful resource for MTBs and data researchers. The development of DrugOn demonstrates the importance of integrating pharmacological data with genomic information to explore and identify potential treatment options. Copy number variations (CNVs) are changes in the number of copies of genes or genomic regions. They play a critical role in the development and progression of cancer. Therefore, the PanCNV-Explorer has been implemented to contribute to pan-cancer analysis. This database provides a comprehensive resource for researchers and clinicians to identify CNVs associated with different types of cancer in many tissue types. It is a powerful resource for the identification of novel genetic alterations in a wide range of cancers. Onkopus, a modular biomarker interpretation framework, enables automated interpretation of genetic variants and analysis of mutations. By combining the above resources and other tools into a single framework with an easy-to-use interface, Onkopus helps clinicians to quickly assess the clinical relevance of genetic variation and make evidence-based treatment decisions. This tool reduces the effort required to interpret large genomic datasets and increases the efficiency and accuracy of clinical decision-making in precision oncology. The application of these tools is demonstrated in a case study focused on mature T-cell lymphoma, a rare and aggressive form of cancer. By analysing patient data with the developed tools, novel therapeutic strategies are identified that show promise for improving treatment outcomes. In conclusion, this thesis provides a valuable contribution to the field of precision oncology by developing tools that address critical challenges in data processing, normalisation and interpretation. SeqCAT, VUS-Predict, DrugOn and the PanCNV-Explorer enhance the ability of MTBs to make informed decisions regarding cancer treatment. In addition, the development of Onkopus highlights the importance of creating user-friendly platforms that simplify the integration of genomic data into clinical workflows. The interdisciplinary nature of this research, combining bioinformatics, oncology and clinical practice, underlines the importance of collaboration in advancing precision medicine. Together, they represent a major step forward in the effort to personalise cancer care and improve patient outcomes.
Keywords: Precision Medicine; Knowledge Management; Variants of Unknown Significance; Data Harmonisation; Drug Ontology; Copy Number Variation; Variant Interpretation; Onkopus; DrugOn; SeqCAT; T-Cell Lymphoma; VUS-Predict; PanCNV-Explorer
