Specifically designed for Stereo-seq platform, GEM_toolkit is a collection of scrips that cooperates with ImageJ/TrakEM2 and CellProfiler to semi-automatically preprocess single-cell or spatially resolved transtriptomics data in 2D/3D using the gene expression matrix (GEM) and ssDNA image.
GEM_toolkit also provides several handly tools for file format conversion, image or data subset, color-code gene heatmap or ssDNA image masking, ROI extraction, affine coordinate calculation, and GEM or other image visualization. Enjoy with GEM_toolkit.
Overview of this workflow :
- pandas
- numpy
- skimage
- scipy
- json
./GEM_toolkit.py -h
Usage : GEM_toolkit.py action [options]
Actions:
---------------------------------------------------------------------
Workflow for single slice to generate single-cell resolved data:
prepare_registration_heatmap generate 8bit spot-level heatmap with highlighted tracklines.
prepare_registration_ssdna generate 8bit close-spot-level ssDNA iamge with highlighted tracklines.
second_registration second round registration.
gem_to_gemc convert GEM into GEMC based on cellbin result and registration results.
Workflow for multiply slices (3D mode) to generate 3D resolved coordinates:
prepare_alignment_image generate 8bit spot-level binary/annatation image for 3D alignment.
apply_alignment set 3D coordinate for GEM(C)/h5ad/ssDNA/cell.mask.
Format coverting tools:
gem_to_h5ad convert GEM to h5ad by a certain binsize.
gemc_to_h5ad convert GEMC to h5ad.
Affine tools:
affine_gem modify the 2D coordinate in GEM(C) by user-defined affine matrix.
affine_h5ad modify the 2D coordinate in GEM(C) by user-defined affine matrix.
affine_ssdna affine the ssdna image based on user-defined affine matrix.
affine_txt affine txt like cell.mask based on user-defined affine matrix.
Region of interest(ROI) tools:
chop_image chop region of interests from whole image.
chop_gem chop region of interests from GEM(C).
Mask tools:
mask_gem mask GEM(C) by mask image.
mask_h5ad mask h5ad data by mask image.
Visualization tools:
draw_heatmap draw heatmap of expression counts in bin1 resolution with/without cellbin and with/without ssDNA.
image_blend merge heatmap/annotation image with ssDNA and border image
Other tools:
chop_paste chop or paste ssDNA image. This tools is useful for ultra-large ssDNA image.
trakEM2_to_affine covert trakEM2_matrix to standard affine matrix.
get_xml_matrix get matrix from trakEM2 xml file.
split_gem split gem by x or y coordinate.
merge_h5ad merge files of h5ad.
gem_xy get xmin ymin of gem
-----------------------------------------------------------------
-h/--help show this short usage
We renamed some action names in this new version and keep the old action names valid for backward compatibility:
prepareregistrationheatmap == prepare_registration_heatmap
prepareregistrationssdna == prepare_registration_ssdna
secondregistration == second_registration
gem_to_cfm == gem_to_gemc
Please cite our GitHub url: http://github.com/BGI-Qingdao/GEM3D_toolkit directly.
The registration scripts have no update between dev branch and the old main branch. However, if you need a software version, then please cite the zendo doi 
The Stereo-seq platform: Chen, et al., Cell, 2022 Spatiotemporal transcriptomic atlas of mouse organogenesis using DNA nanoball-patterned arrays
ImageJ/TrakEM2: Cardona1, et al., PLOS ONE, 2012 TrakEM2 Software for Neural Circuit Reconstruction
CellProfiler: DR, et al., BMC Bioinformatics, 2021 CellProfiler 4: improvements in speed, utility and usability