Department of Computer Science - News


Defence Linus Backert

Wednesday, November 14, at 1:00 pm in room C215, Sand 14

Applied immunoinformatics:
HLA peptidome analysis for cancer immunotherapy

Reviewer 1: Prof. Dr. Oliver Kohlbacher

Reviewer 2: Prof. Dr. Stefan Stevanović

Short summary of the presentation:

Despite billions spent on research and a multiple of established therapeutic approaches, cancer remains one of the leading causes of death world-wide. Therefore, new therapies such as immunotherapy are explored to improve therapeutic options for cancer patients. Immunotherapies need targets that the immune system can use to recognize malignant cells. In our case, these targets are peptides presented by the human leukocyte antigen (HLA), which is a cell-surface protein. Both the malignant and the benign HLA-presented peptides, which together constitute theimmunopeptidome, have to be examined, to define these new targets. However, most published datasets contain only cancer samples or benign samples taken from cancer patients, which might be altered. Therefore, there is a lack of benign tissue samples to establish the benign immupeptidome. To close this gap of data, we collected a large immunopeptidome dataset of benign tissues containing multiple tissue types from different individuals, which is made available as the HLA Ligand Atlas.

The HLA Ligand Atlas is accessible through a web-interface that we developed to accompany the data. It provides a user-friendly option to access the immunopeptidome dataset and a fast, interactive search option which can be used to search for tissue specific HLA-peptides. Furthermore, common statistics are accessible to the user. The HLA Ligand Atlas uses a MySQL database and Pyramid, a Python-based web interface.

Using the large dataset of benign samples, we were able to define general properties of the immunopeptidome. We analysed both the inter- and the intra-individual variability of the immunopeptidome on protein and peptide level. Furthermore, this first large dataset of benign tissue samples allows us to assess properties like the length distribution of different HLA alleles and the nestedness of the peptides in the two HLA classes. Finally, we will demonstrate how quality of the data was ensured and discuss general problems which could occur during data generation.