SPECK: an unsupervised learning approach for cell surface receptor abundance estimation for single-cell...

Area of Science:

• Single-cell transcriptomics
• Computational biology
• Proteomics

Background:

• Single-cell RNA sequencing (scRNA-seq) enables detailed analysis of complex tissues by profiling individual cells.
• Accurate estimation of cell surface protein abundance is crucial for understanding tissue function and cell interactions.
• Existing protein quantification methods are limited by antibody availability and may not be suitable for all tissues.

Purpose of the Study:

• To develop a novel unsupervised method for estimating cell surface protein abundance using scRNA-seq data.
• To evaluate the performance of the new method against existing unsupervised approaches for receptor abundance estimation.

Main Methods:

• Developed SPECK (Surface Protein abundance Estimation using CKmeans-based clustered thresholding), an unsupervised method for receptor abundance estimation from scRNA-seq data.
• Evaluated SPECK's performance using scRNA-seq data for at least 25 human receptors across multiple tissue types.
• Compared SPECK against other unsupervised methods, focusing on techniques based on thresholded reduced rank reconstruction.

Main Results:

• SPECK demonstrated superior performance in estimating cell surface protein abundance compared to other unsupervised methods.
• Thresholded reduced rank reconstruction techniques are effective for receptor abundance estimation from scRNA-seq data.
• The method was validated across various human receptors and tissue types.

Conclusions:

• SPECK provides an effective unsupervised approach for estimating cell surface protein abundance when direct protein measurements are unavailable.
• The method enhances the utility of scRNA-seq data for studying cell surface proteins in diverse biological contexts.
• SPECK is freely available, facilitating its adoption in the research community.