HDV infection shapes a distinct liver microenvironment that is partially restored after viral clearance

This repository contains the analytical pipelines and custom scripts for the single-cell and spatial transcriptomics profiling of HBV and HDV infection in human liver samples. Data integration relies heavily on NanoString CosMx SMI and scRNA-seq spatial mapping.

Highlights

• Microenvironment Remodeling: HDV induces distinct hepatic microenvironment remodeling at both transcriptomic and spatial levels compared to HBV monoinfection.
• Viral Antigen Niches: HBsAg and HDAg establish distinct, highly localized hepatic microenvironments.
• Cellular Communication: Active viral infection systematically disrupts hepatocyte cellular communication, particularly with Kupffer cells.
• Viral Clearance: HDV clearance restores hepatocyte programming, tissue architecture, and cellular interaction networks.

Graphical Abstract

Interactive Data Explorer

To facilitate data exploration and open science, we have deployed a public interactive Shiny web application accompanying this manuscript: CosMx Liver Data Explorer

---

Methods Overview

• QC & Cell Typing: Cells underwent stringent count, probe, and area-based filtering. Supervised annotation into 28 refined cell types was performed using InSituType (mapped against our scRNA-seq reference).
• Neighborhood Analysis: Local cellular niches were defined using proximity-weighted kNN graphs (k=40) and Leiden clustering (resolution = 0.2). Niche biological interpretation was manually curated with LLM assistance.
• DEGs & Pathways: Differential expression was calculated via Seurat's FindMarkers, with pathway enrichment mapped to MSigDB via clusterProfiler.
• Spatial Interactions: Ligand-receptor mappings were modeled using SCOTIA. Spatial co-localization was quantified via Squidpy using a custom Spatial Interaction Enrichment Score (SIES) compared against 1,000 random permutations.
• Pseudotime Trajectories: Transitions between spatial niches were modeled using Slingshot, and dynamically expressed genes along these paths were fitted with General Additive Models (GAMs) using tradeSeq.

Core Dependencies

Ensure your environment matches the following core package versions before executing the analytical pipeline:

• Seurat (v5.4)
• InSituType (v2.0)
• Scanpy (v1.10)
• scikit-learn (v1.3)
• Squidpy (v1.2.3)
• clusterProfiler (v4.10.1)

---

Repository Structure

The repository is organized functionally into data processing arrays (analysis/) and visualization generation (figures/).

.
├── analysis/
│   ├── cosmx/                          # Spatial transcriptomics analytical pipeline
│   │   ├── 1.qc.R
│   │   ├── 2.celltype.R
│   │   ├── 3.cellinteraction.R
│   │   ├── 4.neighborhoods.py
│   │   ├── 5.pseudotime.R
│   │   ├── 6.colocalization.py
│   │   ├── 7.localcomposition.py
│   │   └── scotia_run.py
│   └── single-cell/                    # scRNA-seq processing and integration
│       ├── 01.Cells_together.R
│       ├── 02.Subsets.R
│       ├── 03.Myeloids.R
│       ├── 04.Tcells.R
│       ├── 05.Plasma_and_bcells.R
│       ├── 06.Hepatocytes.R
│       ├── 07.Parenchymal.R
│       ├── 08.Annotation.R
│       ├── 09.All_annotated_together.R
│       ├── 10.cell_communication_analysis.py
│       └── 11.tensor_cell2cell.ipynb
├── figures/                            # Figure generation scripts
│   ├── curate_annotation_cosmx.R
│   ├── fig2.R
│   ├── fig3.R
│   ├── fig4.R
│   ├── fig5.R
│   ├── fig6.R
│   ├── fig7.R
│   ├── supfig2.R
│   ├── supfig3.R
│   ├── supfig5.R
│   ├── supfig6.R
│   └── outs/                           # Output directory for generated plots
├── Graphical_Abstract.png
└── README.md

Usage

1. Download data objects from the Zenodo repository.
2. Execute the processing scripts within analysis/single-cell/ and analysis/cosmx/ sequentially to reproduce the data structures.
3. Run the scripts within the figures/ directory to generate the final manuscript panels. Outputs will be routed directly to the figures/outs/ directory.

Citation

If you use this code, data, or interactive application in your research, please cite our publication:

To be published...

License

This project is licensed under the MIT License - see the LICENSE file for details.