PANCOSMOS
The Pancreatic Cancer COSMOS Atlas — Pancreatic Ductal Adenocarcinoma

Key Features

1

Integrated Multi-Center Atlas

Unified single-cell and spatial data from multiple international PDAC cohorts.

2

Million-Cell Landscape

Over one million single cells defining PDAC tumor microenvironment heterogeneity.

3

AI-Ready Resource

High-quality dataset optimized for virtual cell modeling and AI training.

4

Transferable Cell Annotation

CellTypist-based transfer learning for consistent, hierarchical labels across datasets.

5

Gene Expression Explorer

Query and visualize gene expression across single-cell and spatial datasets.

6

Spatial Colocalization Network

Map spatial organization and cell–cell proximity interactions within tissues.

7

Survival Analysis Module

Gene- and gene-set–level survival modeling across 11 bulk cohorts.

8

FAIR Data Access

Standardized, accessible, and citable data following FAIR principles.

Survival Analysis

🔒
Downloads are locked
Data downloads will be made available upon online publication of the manuscript.

Download Center

Single-Cell

Integrated level-4 single-cell RDS objects (lineage-specific). Click to download:

Stereo-seq level3 bin100 Seurat objects

Processed Seurat objects (bin100). Click to download:

Stereo-seq (bin1)

Raw GEM files for PUMCH (bin = 1). Click to download:

CellTypist cell type prediction models

Pre-trained models for lineage-level prediction. Click to download:

python
                  import celltypist

                  pred = celltypist.annotate(adata, model='PDAC_TME_celltype_Level1.pkl')

                  pred.predicted_labels

PANCOSMOS Tutorial