Abstract

Cutaneous melanoma arises from melanocytes and grows inside skin that already contains keratinocytes, fibroblasts, blood vessels, and both resident and recruited immune cells. Where those populations sit relative to the tumour carries information that a dissociated measurement throws away: immune cells at a tumour margin behave differently from immune cells excluded from it, and the spatial arrangement of myeloid and lymphoid populations in primary melanoma changes as a lesion progresses. Spatial transcriptomics (ST) records gene expression together with the position of each measurement, so expression can be compared with the tissue image and with neighbouring cells, which is what makes a tumour microenvironment (TME) accessible to analysis.

About This Material

This is a Hands-on Tutorial from the GTN which is usable either for individual self-study, or as a teaching material in a classroom.

Questions this will address

• How is a Xenium output bundle turned into a SpatialData object in Galaxy, and what does that object contain?
• Which quality control metrics are specific to segmented cells, and how do they change the filtering decisions?
• How are Leiden groups turned into annotated cell populations, and what evidence is needed?
• What can spatial statistics and ligand-receptor rankings tell us about a tumour microenvironment, and what can they not?

Learning Objectives

• Build a SpatialData object from a Xenium output bundle with SpatialData IO
• Export the expression table to AnnData and inspect its dimensions
• Evaluate transcript-based and morphology-based quality control metrics and map them onto the tissue image
• Apply filters and record how many cells each one removes
• Execute normalisation, feature selection, dimensionality reduction and clustering
• Identify marker genes for each cluster and assign a biological description supported by published evidence
• Appraise CellTypist, Squidpy and LIANA outputs and explain what each one does not establish