ᛅᛋᛏᚱᛁᛏ (ASTRID - Automatized Single-cell Typing for tRanscrIptomics Data)

By Alper Eroglu <alper.eroglu at scilifelab.se>, Jean Hausser <jean.hausser at scilifelab.se>

Tool for automatic annotation of cells from scRNA-seq (Single Cell RNA sequencing) and Xenium in situ sequencing datasets. Designed to work well even with tumor samples. Currently takes AnnData objects from Scanpy as an input.

Run ASTRID (Automatized Single-cell Typing for tumoR transcrIptomics Data) ᛅᛋᛏᚱᛁᛏ pipeline

options:
  -h, --help            show this help message and exit
  --all                 Run all tasks
  --clustering          Run clustering
  --annotation          Run annotation
  --validation          Run validation
  --damage              Run cancer detection from chromosomal damage
  --input_file INPUT_FILE
                        Input file path (/your/input/folder/file.h5ad)
  --input_prefix INPUT_PREFIX
                        Input prefix (Sample0)
  --output_file OUTPUT_FILE
                        Output file path (/your/output/folder/file.h5ad)
  --output_clustering_results OUTPUT_CLUSTERING_RESULTS
                        Output clustering results path (/your/output/folder/astrid_output_file.csv)
  --final_key FINAL_KEY
                        Key for final level of clustering (ASTRID_Clusters) (column in AnnData.obs) 
  --author_type AUTHOR_TYPE
                        Author cell type column name (column in AnnData.obs)

Example bash script in RunASTRID_Ji.sh.

Requirements

• Python (tested in version Python 3.10.10)

  • Numpy - 1.23.4
  • Pandas - 2.2.2
  • Scanpy - 1.9.3
  • scikit-learn - 1.5.0
  • scipy - 1.8.1
  • seaborn - 0.12.2
  • leidenalg - 0.9.1
  • matplotlib - 3.7.2
  • regex
  • infercnvpy - 0.4.3
  • colorir - 2.0.0
  • umap-learn 0.5.3
  • adpbulk - 0.1.3

• R (tested in R version 4.3.3)

  • SingleR - 2.4.1
  • tidyverse - 2.0.0
  • Matrix - 1.6-5
  • SingleCellExperiment - 1.24.0