A Versatile Deep Graph Contrastive Learning Framework for Single-cell Proteomics Embedding

Abstract The advance of single-cell proteomics sequencing technology sheds light on the research in revealing the protein-protein interactions, the post-translational modifications, and the proteoform dynamics of proteins in a cell. However, the uncertainty estimation for peptide quantification, data missingness, severe batch effects and high noise hinder the analysis of single-cell proteomic data. It is a significant challenge to solve this set of tangled problems together, where existing methods tailored for single-cell transcriptome do not address. Here, we proposed a novel versatile framework scPROTEIN, composed of peptide uncertainty estimation based on a multi-task heteroscedastic regression model and cell embedding learning based on graph contrastive learning designed for single-cell proteomic data analysis. scPROTEIN estimated the uncertainty of peptide quantification, denoised the protein data, removed batch effects and encoded single-cell proteomic-specific embeddings in a unified framework. We demonstrate that our method is efficient for cell clustering, batch correction, cell-type annotation and clinical analysis. Furthermore, our method can be easily plugged into single-cell resolved spatial proteomic data, laying the foundation for encoding spatial proteomic data for tumor microenvironment analysis.