Complex Biomarker derivation using genome graphs in cancer

• Applicant: Professor Kjong Lehmann, Ph.D.
• Subject Area: Medical Informatics and Medical Bioinformatics
• Term: since 2026
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 576500477

Every year, thousands of patient genomes are analysed by molecular tumour boards in Germany. However, the factors influencing the outcomes of these workflows are often uncertain. Therefore, it is crucial to identify and understand those factors that influence workflow results and adapt processes to ensure high-quality and equitable patient care at each site. We aim to systematically uncover the influence that genetic diversity may have on the calculation of complex biomarkers and provide new recommendations towards workflow composition. Considering the genetic background is an important aspect in biomedical research, particularly in the study of complex genetic and transcriptomic biomarkers. For this reason, eQTL models typically take germline information into account since differences in genetic background can influence disease prevalence, progression, and response to treatment. Differences in genetic diversity among populations can affect biomarker expression and the efficacy of therapeutic interventions. Recognizing and accounting for these differences ensures that research findings are applicable to a broader population, thereby enhancing the precision and effectiveness of medical interventions. We will systematically collect and organize all relevant data required for this study from the International Cancer Genome Consortium (ICGC) cohort, and The Cancer Genome Atlas (TCGA) cohort. A systematic comparison of the computation of selected common biomarkers using a graph-based pangenome alignment strategy will follow. Finally, we will assess the effect of pangenome alignments compared to linear reference sequence use in the computation of transcriptomic biomarkers. This analysis strategy will enable us to investigate the effect of genomic diversity on biomarker computation and potential mitigating solutions (e.g. pangenome alignment). Identifying potential problematic but also mitigating effects will ensure the accuracy, reliability, and reproducibility of biomarkers for cancer diagnosis and treatment, ultimately enhancing personalized medicine and improving patient outcomes. We believe that this work will support the discovery of more representative and applicable biomarker computation strategies that are applicable to all segments of the population.

• DFG Programme: Research Grants
• Co-Investigator: Dr. Nadina Ortiz-Brüchle