Digital Tissue Deconvolution - Learning from single cells

• Applicant: Professor Michael Altenbuchinger, Ph.D.
• Subject Area: Medical Informatics and Medical Bioinformatics ; Bioinformatics and Theoretical Biology
• Term: from 2019 to 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 420069742

• Final Report Year: 2024

Final Report Abstract

Estimates of cellular compositions increasingly complement analyses of omics data. Bulk transcriptomic profiles typically contain contributions from multiple cells, meaning that the measured gene expression profiles are averages of cellular populations in a tissue. To address this issue, cell-type deconvolution methods were developed. These refer to a class of algorithms to infer cellular compositions from omics profiles. They solve the following inverse problem: Given the expression profile y of a tissue, what is the cellular composition c of that tissue? If X is a matrix whose columns are reference profiles of individual cell types, the composition c can be computed by minimizing L(y − Xc) for a given loss function L. The goal of this research project was to improve the state-of-the-art methodology in the cell-type deconvolution of bulk transcriptomic data. As such, we proposed a new approach named Adaptive Digital Tissue Deconvolution (ADTD), which addressed two major obstacles of cell-type deconvolution: (1) ADTD accounts for potential background contributions of unknown cellular populations, and (2) addresses the molecular adaptation of cells to their tissue. These features make ADTD highly competitive and flexible to adapt to new applications. Moreover, by modelling cellular adaptation, it facilitates estimates of cell-type specific gene regulation from bulk transcriptomic data, setting it apart from established approaches.

Publications

• SpaCeNet: Spatial Cellular Networks from omics data.
  Schrod, Stefan; Lück, Niklas; Lohmayer, Robert; Solbrig, Stefan; Völkl, Dennis; Wipfler, Tina; Shutta, Katherine H.; Guebila, Marouen Ben; Schäfer, Andreas; Beißbarth, Tim; Zacharias, Helena U.; Oefner, Peter J.; Quackenbush, John & Altenbuchinger, Michael
  
• Virtual Tissue Expression Analysis.
  Simeth, Jakob; Hüttl, Paul; Schön, Marian; Nozari, Zahra; Huttner, Michael; Schmidt, Tobias; Altenbuchinger, Michael & Spang, Rainer
  
• Adaptive digital tissue deconvolution. Bioinformatics, 40(Supplement_1), i100-i109.
  Görtler, Franziska; Mensching-Buhr, Malte; Skaar, Ørjan; Schrod, Stefan; Sterr, Thomas; Schäfer, Andreas; Beißbarth, Tim; Joshi, Anagha; Zacharias, Helena U.; Grellscheid, Sushma Nagaraja & Altenbuchinger, Michael
  
• CODEX: COunterfactual Deep learning for the in silico EXploration of cancer cell line perturbations. Bioinformatics, 40(Supplement_1), i91-i99.
  Schrod, Stefan; Zacharias, Helena U.; Beißbarth, Tim; Hauschild, Anne-Christin & Altenbuchinger, Michael
  
• SpaCeNet: Spatial Cellular Networks from Omics Data. Lecture Notes in Computer Science, 344-347. Springer Nature Switzerland.
  Schrod, Stefan; Lück, Niklas; Lohmayer, Robert; Solbrig, Stefan; Wipfler, Tina; Shutta, Katherine H.; Guebila, Marouen Ben; Schäfer, Andreas; Beißbarth, Tim; Zacharias, Helena U.; Oefner, Peter J.; Quackenbush, John & Altenbuchinger, Michael