Spatial transcriptomics (ST) has transformed our understanding of tissue biology, particularly in cancer, where the tumor microenvironment (TME) plays a crucial role in cancer development and progression. However, existing methods often struggle to fully exploit the rich spatial information available in ST data. A key challenge is identifying and analyzing cellular neighborhoods—groups of spatially adjacent cells that interact within the TME. These neighborhoods are essential for studying biological processes, disease mechanisms, and cell-cell interactions, but current tools often fall short in detecting mixed-cell neighborhoods or generating detailed, cell-level profiles.
To overcome these limitations, we developed hoodscanR, a Bioconductor R package designed for comprehensive neighborhood analysis in spatial transcriptomics. hoodscanR efficiently identifies cellular neighborhoods using a k-nearest neighbor search and a hyperparameter-tuned algorithm that generates neighborhood probability distributions at the single-cell level, enabling a refined analysis of spatial patterns.
We applied hoodscanR to breast cancer (10X Genomics Xenium) and non-small cell lung cancer (Nanostring CosMx) datasets, where it successfully detected mixed neighborhoods composed of multiple cell types. For example, in breast cancer, hoodscanR identified spatially adjacent DCIS grade 2 cancer cells and ACTA2+ myoepithelial cells, while in lung cancer, it revealed mixed neighborhoods of B cells and plasma cells. These findings underscore the tool’s ability to capture complex cellular compositions within the TME.
In addition to neighborhood identification, hoodscanR supports downstream analyses such as neighborhood-based differential gene expression, co-localization, and unsupervised clustering. It is computationally efficient, outperforming existing methods in both co-localization and spatial domain identification.
In conclusion, hoodscanR is a powerful and versatile tool that allows researchers to explore cellular neighborhoods with greater precision and depth. Its ability to reveal the spatial context of cellular interactions makes it an invaluable asset in cancer research, facilitating deeper insights into tissue biology and accelerating discoveries in biomedical science.