Dragana Dudić1*, Diana Domanska2, Nicolina Sciarffa3, Francisca Hofman-Vega4, Serafina Reif5 and Nico Trummer5
1 Faculty of Computer Science and Informatics, University Union Nikola Tesla, Belgrade, Serbia
2 Department of Pathology, University Hospital, Oslo, Norway
3 Advanced Data Analysis Group, Ri.MED Foundation, Palermo, Italy
4 Department of Otorhinolaryngology, University Hospital Essen, Essen, Germany
5 TUM School of Life Sciences, Technical University of Munich, Freising, Germany
ddudic [at] unionnikolatesla.edu.rs
Abstract
Head and neck cancer is the seventh most common cancer in the world with squamous cell carcinoma as the most common histology. This heterogeneous group of tumors with aggressive malignancy is characterized by a specific tumor microenvironment in which myeloid cells dominate with its role to initiate antitumor response.
In order to reveal principles of immunity for different regions affected with head and neck cancer, we are creating the atlas of head and neck myeloid cells based on publicly available and in-house single-cell transcriptome datasets created with different single-cell transcriptome sequencing platforms. Current version of head and neck myeloid cells atlas is comprised of core atlas and extended atlas, where the core atlas includes 14 datasets with 102 patients and 582384 cells, created with 10x Genomics sequencing platform, while extended atlas includes 3 datasets with 17 patients and 131380 cells, created with other single-cell sequencing methods like BD Rhapsody, SmartSeq2 and In-Drop.
Dataset specific preparation has been done in R and Python. In order to automate the process for the rest of the analysis, we created a SIMBA pipeline which is publicly available on GitHub. We performed automatic and manual annotation of the head and neck myeloid cells atlas. Automatic annotation is conducted in two ways, using SingleR with multiple datasets and Celltypist with majority voting. Manual annotation is based on markers defined by domain experts. All obtained annotations are publicly available through CellxGene platform.
Our future work includes expanding the head and neck myeloid cell atlas with novel datasets, deconvolution based on bulk transcriptome data available in TCGA database and addition of spatial transcriptomics data.
Keywords: bioinformatics, single cell transcriptomics, scRNA-seq, head and neck cancer
Acknowledgement: This publication is based upon work from COST Action Mye-InfoBank CA20117, supported by COST (European Cooperation in Science and Technology).