UMscAtlas

Analysis of healthy, primary, and metastatic uveal melanoma using scRNA-seq and Xenium spatial transcriptomics. This repository includes workflows for preprocessing, integration, cell-type annotation, and characterization of malignant and TME states.

Metadata variables documentation

Metadata variable	Name in figure	Explanation
primary_location	Intraocular Location	Location within the eye where the primary tumor arose. NA for samples that are not primary UM
origin	Disease Progression	Indicates if cells come from a healthy sample, primary tumor or metastasis
orig.ident	Sample	Identifies physical tissue sample cells originate from
location	Location	Location (intraocular location and location of metastasis) for all samples in the data set
samplename	Sample Name	Identifies biological replicates sequenced in different libraries (use instead of Sample column (that shows projectnumber and Index from FGCZ
healthy_location	Intraocular Location Healthy	Identifies intraocular location from healthy tissue
Condition	Batch	Indicates sequencing batch for each sample
primary_mutation	Primary Mutation	Identifies primary mutation (GNAQ, GNA11, PLCB4 or CYSLTR2) for each sample (assessed by WES or targeted sequencing)
secondary_mutation	Secondary Mutation	Identifies secondary mutation (BAP1, SF3B1 or EIF1AX) for each sample (assessed by WES or targeted sequencing)
Patient_nr	Patient Number	assigns a patient number (P1-P36) to each sequenced sample and shows which patients have multiple samples sequenced
organ	Organ	Identifies organ from which sample is originating
Gender	Gender	Identifies gender of patient sample was derived from
TreatmentStage_ofProcessedSample	Treatment	Indicates last treatment of patient before sampling of the sequenced sample
Stage_atDiagnosis	Stage at Diagnosis	Disease stage at first diagnosis
Mets_atDiagnosis	Metastasis at Diagnosis	Identifies if metastasis was present at disease prognosis
Age_atDiagnosisY	Age at Diagnosis	Identifies age at diagnosis of tumor in years
RiskCategory	Risk Category	Identifies risk category assigned by collaborators in Australia for primary tumors
Time_sinceDiagnosis	Time since Diagnosis	Shows calculated time since diagnosis in years until October 2025
PrimaryTreatment	Primary Treatment	Identifies treatment of primary tumor
LocalProgressionPrimary	Local Progression Primary Tumor	Indicates wheter primary tumor showed disease progression or not
LocalProgressionTreatment	Treatment of Local Progression	Identifies which treatment was used if didease was locally progressive
DevelopmentOfMetastasis	Metastasis Development	Indicates whether disease developed into metastatic disease from primary tumor
Time_Primary-Met.Y	Time to Metastasis	Indicates time in years from primary tumor to metastasis development
MetastaticSite	Metastatic Site	Identifies site of metastasis development
MetastasisTreatment	Treatment Metastasis	Treatment of metastatic tumor
Time_sinceMetastasisDiagnosisY	Time since Diagnosis of Metastasis	Identifies time since diagnosis of metastasis in years until october 2025
PatientStatus	Status Patient	Indicates whether patient is deceased or alive
LocationPrimaryTumor	Location of Primary Tumor	Identifies intaocular location of primary tumor also for metastases samples

Symbols to Use in Plots for h-p-m

• healthy: ● Circle (nr. 1)
• primary: ▲ Triangle (nr. 2)
• metastases: ◆ Diamond (nr. 5)

Symbols to Use in Plots for primary mutation

• GNAQ: ● Circle (nr. 1)
• GNA11: ▲ Triangle (nr. 2)
• PLCB4: ◆ Diamond (nr. 5)
• none: ▼ Triangle, point down (nr. 6)
• unknown: ✵ Star (nr. 8)

Available Datasets

• Full Dataset: All cells across all samples.
• TME Subset: Tumor microenvironment cells, subsetted from Full Dataset (all cells except Melanocytes and Melanoma cells).
• Melanoma/Melanocytes Subset: Melanoma cells and melanocytes from healthy - primary - metastases subsetted from Full Dataset.
• Melanoma Subset: Subset of melanoma cells from p-m UM samples subsetted from Full Dataset.

Notes:

• Melanoma cells were defined using inferred copy number variation (iCNV) profiles to distinguish malignant cells from non-malignant Melanocytes populations.
• All subsets are Harmony-corrected, except the Full Dataset.

graph TD
  A[<b>Full Dataset</b><br/>13_final_Dataset_reclustered_28012025.qs<br/>*not integrated*] --> B[<b>TME Subset</b><br/>01_integrated_multicorr_TME.qs<br/>*harmony corrected*]
  A --> C[<b>Melanoma/Melanocytes Subset</b><br/>01_integrated_melanoma_melanocyte_subset.qs<br/>*harmony corrected*]
  B --> E[<b>T/NK cell Subset</b><br/>01_integrated_TNK_subset.qs<br/>*harmony corrected*]
  C --> D[<b>Melanoma Subset</b><br/>01_integrated_melanoma_subset.qs<br/>*harmony corrected*]

Loading

Metadata Structure in Objects

Each Seurat object contains the following components:

1. meta.data

Cell-level annotations and sample information:

• majority_celltype: Broad cell type annotation.
• SubTyping: Detailed immune cell subtypes.
• orig.ident: Original sample identifier (use for all analyses except GloScope).
• samplename: Library identifier.

Tip: Refer to the summary table at the top and the provided R script for recommended naming conventions and color palettes.

---

2. reductions

Dimensionality reduction results for visualization and integration:

• umap: UMAP coordinates for clustering and visualization.
• pca: PCA embeddings for exploratory analysis.
• Integration: All subsets are Harmony-corrected (except the Full Dataset).
• use reduction: umap for Full Dataset
• use reduction: umap_Condition_and_orig.ident_50PC_theta2 for Harmony-corrected Datasets

---

3. assays

Gene expression data:

• RNA: Normalized gene expression matrix (default assay for most analyses).