Spatial analysis playbook

Author

Alex Qin, Shreya Rao, Farhan Ameen, Ellis Patrick

Published

Invalid Date

Introduction

Welcome!

Recent advances in highly multiplexed cell imaging technologies—such as PhenoCycler, IMC, CosMx, Xenium, and MERFISH—have fundamentally transformed our ability to study complex cellular relationships within tissues. While traditional immunohistochemistry protocols were limited to visualising cells based on just two or three surface proteins, these cutting-edge technologies can now characterise cells using over 50 proteins or thousands of RNA molecules in situ. This breakthrough enables precise classification of cell subtypes and offers an unprecedented view of cellular heterogeneity in tissue environments.

To fully leverage the potential of spatial omics, novel analytical approaches are essential—not just new methods, but also a clear understanding of when and where to apply them. This playbook supports that need by aligning analytical strategies with biological questions, tissue structure, and data characteristics. We use packages developed within the scdney ecosystem to illustrate key challenges, trade-offs, and analytical decisions across each stage of the workflow. However, our focus is not on prescribing a single workflow. Instead, we aim to equip researchers with a flexible framework that emphasises interpretation, reproducibility, and the biological insights that emerge from thoughtful, context-aware spatial analysis.

We present a flexible seven-stage analytical workflow designed to accommodate a range of imaging technologies and biological contexts:

  1. Cell segmentation and pre-processing
  2. Quality control and normalisation
  3. Cell clustering/annotation
  4. Quantifying co-localisation between cell types
  5. Identifying spatial domains
  6. Measuring changes in marker expression
  7. Classification of patients to clinical outcomes

Rather than focusing on the specific technologies used, we encourage researchers to centre their analyses on the biological questions they aim to answer. By doing so, this framework helps ensure that analytical choices are guided by scientific goals and contextual knowledge, rather than technical convenience alone.