Skip to content

Latest commit

 

History

History
41 lines (27 loc) · 1.52 KB

README.md

File metadata and controls

41 lines (27 loc) · 1.52 KB
Raw

Benchmark

Datasets

All datasets used in benchmark are public available. Here is a summary:

Dataset Species Technology Cells Genes Tissue Disease URL
1 Mouse MERFISH 378,918 374 Brain Health Link
2 Human Xenium 164,079 313 Breast Tumor Link
3 Human 10X chromium 8,405 31,915 PBMC Health Link

Methods

  • DECIPHER
  • SLAT
  • BANKSY
  • STAGATE
  • scVI
  • Scanpy
  • Hamrony

Reproducibility

Setup environment

Follow the guide to set up the environment. The absolute version of packages is also provided in conda.yaml.

Prepare data

Download the datasets and move them into ./input folder.

Run the workflow in slurm cluster

Take about 12 hours when using 32 A100 GPUs.

We use snakemake to run benchmark workflows in Slurm cluster. Firstly, you need modify partition name in config according to your cluster settings, then run:

snakemake -j 500 --profile profiles/hpc