Back

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

Li, W.; Yang, F.; Wang, F.; Rong, Y.; Wu, B.; Zhang, H.; Yao, J.

2022-12-15 bioinformatics
10.1101/2022.12.14.520366 bioRxiv
Show 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.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.