STEA: Histologically Validated and Reference-Independent Major Cell-Type Annotation for Spatial Transcriptomics Reveals Relevant Cellular Organization and Architecture of Tumor Microenvironment
Open Access
Abstract
Background: Recent advances in spatial transcriptomic technologies enable in situ gene expression profiling while preserving spatial context. This capability is particularly important for studying the tumor microenvironment (TME), where diverse and admixed cell populations interact within highly organized spatial niches that influence tumor progression and therapeutic response. However, the limited resolution of early spatial transcriptomic platforms results in each spatial spot capturing transcripts from multiple cell types, making accurate spot deconvolution or annotation a critical yet challenging step in downstream data analysis. The level of complexity will be particularly prominent in heterogeneous samples like the tumor microenvironments where multiple cell types are highly admixed and reliable single-cell reference atlases may usually be unavailable. Methods: In this paper, we developed our method called STEA, which is a novel and accurate reference-independent enrichment-based annotation algorithm for major cell type. Unlike the existing approaches, STEA does not require single-cell RNA sequencing datasets as reference, offering both flexibility and computational efficiency in execution. Results: We performed comprehensive benchmarking using a variety of simulated datasets across different platforms and scenarios and demonstrated the superior accuracy of STEA. Apart from synthetic data, we also evaluated multiple real datasets to further exemplify its practical applicability on both oncology-related and oncology-unrelated data. More importantly, we could confidently demonstrate the high concordance between prediction of STEA and histological classification by experienced pathologist. Conclusion: Our STEA algorithm provides a practical reference-independent framework to complement the cutting-edge spatial transcriptomics in genomics studies, facilitating accurate downstream high-dimensional spatial characterization of cellular and molecular landscapes, reconstruction of tissue architecture as well as cell–cell communication in malignant and non-malignant scenarios. Taken together, our comprehensive evaluation demonstrates the robustness and reliability of STEA, highlighting its potential as a valuable tool for studying complex tissue organization, particularly within heterogeneous TME.Keywords
Funding Information
- Hong Kong Research Grants Council Theme-Based Research Scheme (T12-716/22-R)
This publication has 40 references indexed in Scilit:
- What is a cell type and how to define it?Cell, 2022
- Reference-free cell type deconvolution of multi-cellular pixel-resolution spatially resolved transcriptomics dataNature Communications, 2022
- Applicability of spatial transcriptional profiling to cancer researchMolecular Cell, 2021
- UCell: Robust and scalable single-cell gene signature scoringComputational and Structural Biotechnology Journal, 2021
- High-Spatial-Resolution Multi-Omics Sequencing via Deterministic Barcoding in TissueCell, 2020
- Spatially Resolved Transcriptomes—Next Generation Tools for Tissue ExplorationBioEssays, 2020
- Tumor Cell Biodiversity Drives Microenvironmental Reprogramming in Liver CancerCancer Cell, 2019
- SCENIC: single-cell regulatory network inference and clusteringNature Methods, 2017
- Systematic RNA interference reveals that oncogenic KRAS-driven cancers require TBK1Nature, 2009
- PGC-1α-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetesNature Genetics, 2003