CITEgeist: Cellular Indexing of Transcriptomes and Epitopes for Guided Exploration of Intrinsic Spatial Trends

Area of Science:

• Genomics
• Biotechnology
• Computational Biology

Background:

• Spatial transcriptomics links gene expression to tissue architecture.
• Current deconvolution methods often require costly single-cell RNA sequencing (scRNA-seq) references.
• Limited tissue samples, like core biopsies, pose challenges for scRNA-seq reference availability.

Purpose of the Study:

• To introduce CITEgeist, a novel tool for deconvoluting spatial transcriptomics data.
• To provide a cost-effective and biologically grounded alternative to scRNA-seq references.
• To enable accurate cell type deconvolution using antibody capture from the same tissue section.

Main Methods:

• Leverages cell surface protein measurements from antibody capture on the same slide.
• Employs mathematical optimization with sparsity constraints to estimate cell type proportions and gene expression.
• Validates CITEgeist using simulated data and clinical samples, including ER+ breast tumors.

Main Results:

• CITEgeist achieves improved accuracy in cell type resolution, especially in dense tumor microenvironments.
• Demonstrates computational efficiency compared to state-of-the-art deconvolution methods.
• Shows robustness and applicability in translational research using clinical samples.

Conclusions:

• CITEgeist offers a scalable, accurate, and reference-independent solution for spatial transcriptomics deconvolution.
• This antibody-based approach circumvents scRNA-seq limitations, making it suitable for limited tissue specimens.
• The tool advances the analysis of complex tissues and has implications for translational studies.