SECTOR: Processed Spatial Transcriptomics Datasets for Benchmarking and Case Studies

Two processed large-scale, high-resolution spatial transcriptomics datasets for benchmarking clustering and inference methods. The repository includes a 10x Visium HD human colorectal cancer dataset and a 10x Xenium human breast cancer dataset, each provided at multiple spatial resolutions. It also contains processed datasets for two case studies and reproducibility code for the SECTOR study.

Use Cases

• Benchmark spatial clustering algorithms based on high-resolution spatial transcriptomics data
• Evaluate spatiotemporal inference methods using datasets with multiple spatial granularities
• Assess method performance on recovering spatially organized pseudotemporal patterns using case study data
• Reproduce analyses from the SECTOR study using provided code and notebooks

Strengths

• Includes two large-scale, high-resolution spatial transcriptomics datasets totaling 2.3 GB
• Datasets are provided in .h5ad format at multiple spatial resolutions (e.g., 8- and 16-μm bins)
• Reproducibility materials include source code, Python scripts, and three tutorial notebooks

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment

Provenance

Source

figshare

Collection Method

Processed from original studies by Oliveira et al. (Nature Genetics, 2025), Sun et al. (bioRxiv, 2025), Bhuva et al. (Genome Biology, 2024), Andersson et al. (Nature Communications, 2021), and Zhang et al. (Nature, 2023).

Freshness

Last updated 2026-05-14 16:30:02; freshness should be verified