Mapping Biology in Space — DBiT-seq and the Basics of Spatial Multi-Omics Analysis

Training materials for the 2026 V4SDB Student Summer School

Instructor:

Shuozhen Bao (KTH Royal Institute of Technology, Stockholm)
shuozhen.bao@scilifelab.se

Workshop Overview

Sequencing tells you what molecules are present in a sample; it rarely tells you where. Bulk sequencing averages an entire tissue into one profile, and single-cell sequencing resolves individual cells only by dissociating the tissue and destroying their spatial context. Spatial multi-omics closes that gap: it keeps each molecular readout pinned to its original (x, y) position in the tissue.

This session introduces spatial multi-omics analysis through DBiT-seq (Deterministic Barcoding in Tissue), a microfluidic method that stamps a barcode grid directly onto a tissue section. Two perpendicular sets of channels tag every pixel with a unique barcode combination, so RNA, chromatin accessibility (ATAC), and protein (via antibody-derived tags) can all be measured on the same pixels of the same section — three molecular layers, one spatial grid.

Participants work hands-on in a browser-based notebook (no installation required) with a real published dataset: a P21 mouse brain profiled by spatial ARP-seq (ATAC + RNA + Protein) from Zhang et al., Nature 647, 213–227 (2025). No programming background is assumed — most cells are pre-written and simply run and read; a handful of “Your turn” cells invite participants to change one value and observe the effect.

What the session covers

  • Why spatial multi-omics — how it differs from bulk and single-cell sequencing, and where DBiT-seq sits relative to other platforms (e.g., Visium HD, spatial ATAC-seq).
  • Reading the figures — how to interpret the chart types common to omics papers: bar plots, heatmaps, dot plots, and GO/KEGG enrichment as bubble plots or networks.
  • The tissue and the study — mouse cortical anatomy at postnatal day 21 (layered neurons, the myelinating corpus callosum) and what the source paper found: cortical layers sharpen and the corpus callosum myelinates over development, with chromatin accessibility changes preceding transcriptional changes at the same loci.
  • Hands-on analysis — loading and aligning the three molecular tables to a shared pixel grid, checking tissue orientation, visualizing RNA/ATAC/protein signal for the same gene, handling sparse ATAC signal by aggregating gene programmes, and normalizing each modality.
  • Validation — plotting a panel of known cortical-layer marker genes (Cux2, Rasgrf2, Rorb, Bcl11b, Foxp2, Tle4, Mbp, Plp1) and checking that they land in the expected anatomical layers, as the criterion for trusting the rest of the analysis.

Learning objectives

By the end of the session, participants should be able to:

  1. Explain what spatial multi-omics adds over bulk and single-cell sequencing.
  2. Describe how DBiT-seq co-registers RNA, ATAC, and protein measurements to the same tissue pixels.
  3. Read and interpret the standard chart types used to report omics results.
  4. Load, align, and visualize multi-modal spatial data in a shared coordinate system.
  5. Use known anatomical markers to validate that a spatial dataset is correctly oriented and biologically sound before drawing further conclusions.

Format

Self-paced, browser-based (Google Colab), roughly following five blocks: environment setup, why spatial multi-omics, reading the figures, the tissue and the study, and hands-on analysis with validation. Demonstrators are available throughout for troubleshooting.

Slides

Jupyter notebook: Download it from here

Dataset: Zhang, D. et al. (2025). Spatial dynamics of brain development and neuroinflammation. Nature 647, 213–227. doi:10.1038/s41586-025-09663-y — RNA: GEO GSE308526 · ATAC: GEO GSE308599 · Protein: NeMO Archive.