Uveitis Atlas 

Data accompanying Belean et al. (title coming soon) and the analysis code at https://github.com/BorchLab/Uveitis-Atlas.

This archive holds the core data products for a paired eye (intraocular fluid) and blood single-cell RNA + adaptive immune receptor (TCR/BCR) study comparing viral and non-infectious uveitis (NIU). The code that produces and consumes every file here lives in the GitHub repository above; this Zenodo record holds only the data, which is too large for GitHub.

The deposit is focused on data delivery: the integrated atlases, the raw CellRanger inputs needed to rebuild them, per-sample QC, and the VDJdb reference. Derived figures and tables are **not** included (they are in the manuscript and regenerate from the atlases via the code), and neither are several large intermediates, see "Deliberately excluded" below.

How to use

1. Clone the code: git clone https://github.com/BorchLab/Uveitis-Atlas
2. Download the archives below and unpack each into the matching directory inside
   the cloned repo (the repo ships empty placeholders that mark each location):

   tar -xf 01_objects.tar -C outputs/          # -> outputs/objects/
   tar -xf 02_qc.tar -C outputs/               # -> outputs/qc/
   tar -xf 03_inputs_raw.tar -C inputs/data/   # -> inputs/data/runs/
   tar -xf 04_references.tar .                 # -> references/

3. Verify integrity: shasum -a 256 -c sha256sums.txt
4. Recreate inputs/data/metadata.csv from the manuscript supplementary table 1.

What to download for what

• Re-use / re-analyze the atlases: 01_objects - the integrated full atlas, eye and per-compartment sub-atlases, the Milo differential-abundance object, and the TCR analysis result objects. Toggle the relevant `steps:` flags to re-run.
• Rebuild from raw counts: 03_inputs_raw plus the metadata table, then enable ingest + integration. UMAP coordinates will differ (embeddings are not seeded); cluster IDs are stable. 
• Regenerate figures/tables: these are not deposited; run the viz/analysis steps from `01_objects` (see `config/config.run.yml`).

Contents

01_objects — analysis objects (`.rds`)

• IntegratedSeuratObject.rds — full immune atlas (~3.5 GB)
• eye/ — eye sub-atlas + myeloid/, bcell/, tcell/ per-compartment sub-atlases (~6.7 GB)
• MiloObject_tissue.rds — Milo differential-abundance object (~5.5 GB)
• TCR result objects: ImmLynxTcrdistResults.rds, tcrdist_pw_beta.rds, ImmGLIPHResults.rds, GliphTcrdistJoint.rds, ImmLynxOlgaResults.rds, VdjdbAnnotation.rds, NovelTcrCandidates.rds

02_qc — per-sample QC reports

Scatter and violin QC PDFs, one set per sample.

03_inputs_raw — raw CellRanger per-sample outputs

Per-sample gene-expression (count/sample_filtered_feature_bc_matrix) and VDJ (vdj_t, vdj_b) outputs. 

04_references — VDJdb cache

references/vdjdb/ snapshot used by TCR annotation (R/69_vdjdb_annotation.R).

Integrity

sha256sums.txt lists a SHA-256 for every archive. Verify with:

shasum -a 256 -c sha256sums.txt

License

Creative Commons Attribution 4.0 International (CC BY 4.0). If you use these data, please cite the paper and this Zenodo record.

Citation

• Paper: coming soon (DOI TBD)
• Code: https://github.com/BorchLab/Uveitis-Atlas
• This archive: 10.5281/zenodo.20783773