Spatial Transcriptomics · ratschlab
TCGA_virtual_spatial_transcriptomics_atlas
A virtual spatial transcriptomics atlas across 32 cancer types from 28,664 whole-slide images in TCGA.
What TCGA_virtual_spatial_transcriptomics_atlas is
The TCGA virtual spatial transcriptomics atlas is a population-scale biological resource created by ratschlab. It provides computationally inferred spatial transcriptomic data generated by augmenting retrospective histopathology archives from The Cancer Genome Atlas (TCGA).
Covering 28,664 whole-slide images from 10,865 patients across 32 different cancer types, the dataset provides researchers with large-scale spatial molecular data paired with tumour pathology.
How it was made
The atlas was constructed using DeepSpot-M, a deep learning system that computationally augments standard whole-slide histopathology images into virtual spatial transcriptomics data.
Key facts
- Generating method
- DeepSpot-M
- Whole-slide images
- 28,664
- Patients
- 10,865
- Cancer types
- 32
- Source cohort
- The Cancer Genome Atlas (TCGA)
What you can use it for
- Cross-cancer spatial transcriptomics analysis
- Examining spatial gene expression patterns across 32 diverse cancer types using computationally augmented histopathology slides.
- Population-scale oncology research
- Investigating tissue heterogeneity and molecular variations across a broad cohort of 10,865 patients.
- Computational pathology model benchmarking
- Developing and evaluating machine learning algorithms that link whole-slide imaging architecture with transcriptomic profiles.
Explore the data
This dataset is gated
Its authors require their terms to be accepted before the data can be read. Accept them on the original dataset page. The platform does not accept terms on your behalf, so the rows are not shown here.Strengths and limits
Good at
- Covers a population-scale cohort of 10,865 patients and 28,664 whole-slide images.
- Encompasses 32 distinct cancer types from the TCGA archives.
Watch out for
- Transcriptomic profiles are computationally inferred rather than experimentally measured.
- Coverage is restricted to retrospective samples archived within TCGA.
Questions
Can I look inside TCGA_virtual_spatial_transcriptomics_atlas without downloading it?
Not here: its authors gate it behind terms each user must accept on the original page. Its description and statistics are still shown.
What licence is TCGA_virtual_spatial_transcriptomics_atlas under?
Its authors declare cc-by-nc-sa-4.0. Read the terms on the original page before reusing the data.
What method generated the virtual spatial transcriptomics data?
The atlas was generated using DeepSpot-M to computationally augment histopathology images.
How many samples and cancer types are included in the atlas?
The dataset contains 28,664 whole-slide images from 10,865 patients across 32 cancer types.
What original data source was used?
The atlas was derived from retrospective histopathology whole-slide image archives in The Cancer Genome Atlas (TCGA).
Written from TCGA_virtual_spatial_transcriptomics_atlas's own dataset card. Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.