Raphael Kirchgaessner

This paper imputes single-cell protein abundance in multiplex tissue imaging data from a breast cancer cohort using machine learning, showing improved accuracy and biological relevance.

This study aggregates multimodal cancer data across unaligned embedding spaces and finds that tumor of origin signal is preserved for learning tasks.

The paper develops a computational strategy to identify gene signatures defining cancer subtypes using multi-omics data, achieving 92% model accuracy and identifying 20 significant genes in breast and glioma cancers.