Workflow to preprocess ST data

1. Collect and download ST data

We usually download ST data from GEO database. Related codes can be found at GEO_ID_download.sh and GEO_list_download.sh.

2. Standardize file format for batch processing

To run this pipeline aotumatically, we need standardize file format.

2.1 Standardize ST gene expression file

ST expression data may have multiple file formats, such as HDF5, plain text, and mtx format. This pipeline mainly uses HDF5 format. Expression files in other formats can be converted into HDF5 files through MAESTRO.

The following commands can be used for format conversion.

MAESTRO mtx-to-h5	      #Convert mtx format to HDF5 format
MAESTRO count-to-h5       #Convert plain text format to HDF5 format

See https://github.com/liulab-dfci/MAESTRO for more information, or enter 'MAESTRO count-to-h5 --help' on the command line to ask for help.

2.2 Standardize spatial image file

To import spatial image information, create a folder named 'sample_spatial'. This folder should contain both images and spot location information. Acceptable image formats include 'tissue_hires_image.png', 'tissue_lowres_image.png', 'sample_HE.tif', 'sample_HE.png' and 'sample_HE.jpg'. Acceptable location information formats include 'tissue_positions_list.csv' and 'sample_spot_location.txt'. Additionally, we can also read in 'scalefactors_json.json' for additional image information if available.

3. Preprocess ST data

R language "Seurat" package is used to preprocess ST data, the specific code is in ST_preprocess.R.