Pathway-centric interpretation of salivary gland remodelling through histology-anchored spatial transcriptomics, single-cell, and bulk RNA

Code: BBSRC-DFA_2026_09

Primary Supervisor: Michael Barnes
Email: m.r.barnes@qmul.ac.uk 
Institute: WHRI

Secondary Supervisor: Greg Slabaugh
Email: g.slabaugh@qmul.ac.uk 
Institute: DERI 

CASE Partner: OracleBio (Only one of the three projects listed with OracleBio will become a CASE project: BBSRC-DFA_2026_09, BBSRC-DFA_2026_25, and BBSRC-DFA_2026_26.)

Abstract:

Human exocrine glands are highly organised secretory tissues in which epithelial, stromal and immune compartments coordinate to maintain secretion and mucosal integrity. The principles governing how these 3 programmes are patterned in space in intact human glands, and how they reconfigure under sustained immune-mediated perturbation, remain poorly defined because bulk transcriptomics averages over microenvironments and single-cell approaches lose spatial context. This PhD will use minor salivary gland (MSG) biopsies as an experimentally accessible human tissue system to define core organisational programmes and their remodelling across a spectrum of immune infiltration, fibrosis, epithelial change and lymphoid neogenesis. Clinical biopsies provide a functional readout of altered exocrine performance, enabling linkage of spatial programmes to gland-level physiology with an emphasis on fundamental tissue organisation and enabling methodology. We will integrate matched bulk RNA-seq, single-cell RNA-seq and histology-anchored spatial transcriptomics. First, we will build a multimodal atlas of MSG microanatomy, mapping cell types/states and transcriptional programmes to ducts, acini, stromal niches, vascular regions, immune aggregates and tertiary lymphoid structure-like areas. Second, we will further develop and validate a generalisable pathway-centric framework (ProtoPathway) to compute pathway activities from bulk and scRNA-seq and project them onto spatial maps enhanced via histology-anchored gene augmentation. Third, we will identify recurring spatial “programme states” and build transferable predictors that infer these states from histology plus bulk or single-cell profiles, enabling scalable application beyond the spatially profiled subset. Deliverables include an MSG atlas-style resource, reproducible software workflows, and robust, interpretable models for multimodal spatial biology with broad applicability across bioscience.

Lay Summary:

Salivary glands are “exocrine” tissues that produce fluids needed for everyday functions such as speaking, eating and protecting the mouth. They are built from different interacting components, secretory and duct cells, blood vessels, supporting structural cells and resident immune cells, arranged in a precise 3D architecture. A major challenge in bioscience is to understand how this tissue organisation is programmed and maintained, and how it adapts when the immune system is persistently activated within a tissue. This PhD uses minor salivary gland biopsies as an experimentally accessible human model to discover general rules of exocrine gland organisation and immune–tissue communication. Samples span a natural range from normal tissue to tissue with chronic immune infiltration and organised immune structures; this provides a powerful “in vivo disruption” to reveal regulatory programmes that are difficult to detect in baseline tissue. Measures of dryness (sicca) provide a functional readout of altered secretory performance, allowing us to relate spatial tissue programmes to gland physiology without shifting the project’s focus away from fundamental biology. The project will integrate three complementary data types from the same biopsies: (1) whole-tissue gene activity profiles, (2) single-cell gene activity profiles that identify distinct cell types and states, and (3) spatial transcriptomics that preserves tissue structure so we can map molecular programmes to ducts, acini, stromal niches, blood vessels and immune aggregates. We will develop and apply new AI methods that combine standard microscope images with spatial and single-cell data to generate higher-resolution, more complete “maps” of tissue programmes. Results will be summarised in interpretable pathway maps, showing which core biological processes are active in specific microenvironments. Outputs will include a reusable computational workflow and an atlas-style resource (within ethical constraints) that will help bioscientists study tissue organisation and spatial regulation across many systems beyond the salivary gland. 

Aims and Objectives:

Aim 1: Build a histology-anchored multimodal atlas of MSG organisation across homeostasis and immune-mediated perturbation 

• Curate matched datasets (bulk RNA-seq, scRNA-seq, spatial transcriptomics, histology) for the MSG biopsy collection, including histological features (e.g., focus score, fibrosis, epithelial remodelling, immune aggregate/TLS presence). 

• Generate/assemble spatial transcriptomic maps aligned to histology, with compartment annotation (acini, ducts, stroma, vasculature, immune aggregates). 

• Quantify how cell-type composition and gene programmes vary across compartments and across perturbation severity. 

Aim 2: Develop and validate a generalisable pathway-centric framework for spatial multi-omics interpretation 

• Extend ProtoPathway to compute pathway activity scores from bulk and scRNA-seq and project these onto spatial maps enhanced by histology-anchored augmentation (benchmarked against similar tools SpatialEx, etc). 

• Benchmark stability and interpretability of pathway maps across donors and serial sections, and identify which cell types drive pathway signals in specific microanatomical niches. 

• Identify spatially localised pathway modules linked to epithelial remodelling, stromal/ECM dynamics, and immune organisation (including TLS-like structures) as generalisable programme states. 

Aim 3: Create transferable predictors that infer spatial pathway states from histology plus omics 

• Train models that predict spatial programme states using histology-derived features and bulk/scRNA-seq summaries. 

• Validate prediction performance in held-out biopsies and quantify which histological cues best explain inferred spatial pathway states. 

• Deliver scalable workflows for applying programme-state inference to additional biopsy collections where spatial profiling is not available.