Spatial Transcriptomics

The foundational mechanism where mRNA from a tissue section is captured onto a surface arrayed with spatially indexed oligonucleotide barcodes. Each spot or bead contains a unique positional barcode that is ligated to captured transcripts during library preparation. This process encodes the physical x,y coordinates of each transcript directly into the sequencing read, allowing computational reconstruction of gene expression maps aligned to the original tissue image. Technologies like 10x Visium use printed spots, while Slide-seq employs randomly deposited beads with in-situ indexing.

A class of imaging-based approaches that directly visualize transcripts within intact tissue without sequencing library extraction. MERFISH (Multiplexed Error-Robust Fluorescence In Situ Hybridization) uses combinatorial barcoding and sequential rounds of hybridization to identify thousands of genes. seqFISH employs sequential barcoding with temporal color coding. These methods achieve subcellular resolution by imaging individual mRNA molecules as diffraction-limited spots, enabling precise quantification and localization within cellular compartments.

A targeted approach where padlock probes hybridize to specific mRNA targets directly in tissue sections. After hybridization, the probes are circularized by ligation and amplified via rolling circle amplification (RCA), creating localized DNA concatemers. These amplicons are then sequenced by synthesis or ligation directly on the tissue, reading out gene-specific barcodes. HybISS and STARmap are prominent variants. ISS offers high specificity due to dual-probe recognition and avoids the optical crowding limitations of hybridization-only methods.

Computational frameworks that combine spatial transcriptomics with complementary data modalities to enrich biological interpretation. Key integration strategies include:

• Image alignment: Registering H&E or immunofluorescence images with expression maps for morphological correlation
• scRNA-seq deconvolution: Using reference single-cell atlases to estimate cell-type proportions at each spatial location via tools like RCTD or cell2location
• Multi-modal fusion: Integrating spatial epigenomics (spatial ATAC-seq) or spatial proteomics (CODEX, MIBI) with transcriptomic layers
• 3D reconstruction: Aligning serial tissue sections to build volumetric expression atlases

Analytical methods that characterize the cellular microenvironment by examining the composition and gene expression of cells within defined spatial neighborhoods. This includes:

• Cell-cell colocalization: Identifying statistically significant spatial relationships between cell types
• Ligand-receptor analysis: Mapping intercellular signaling by detecting co-expression of ligand-receptor pairs in adjacent cells or spots
• Spatial autocorrelation: Measuring how gene expression similarity decays with physical distance using metrics like Moran's I
• Tissue domain identification: Segmenting tissue into functionally coherent regions based on spatially aware clustering algorithms

Statistical frameworks that identify genes whose expression patterns exhibit structured spatial variation rather than random distribution. Methods like SpatialDE and SPARK-X model gene expression as a function of spatial coordinates using Gaussian process regression or generalized linear spatial models. These tools distinguish spatially variable genes from technical noise, revealing genes that define anatomical structures, gradients, or tissue boundaries. This analysis is critical for discovering biomarkers that are not merely cell-type-specific but are dependent on tissue context and microenvironment.