lasso tool

[adilysyn]

I would like to use the lasso tool on StereoMap to pick up cells that belongs only to a specific cluster.
However, since the lasso pointer to define the region is quite large, I also pick adjacent cells which are from different clusters.
If I choose a cluster choose cells by the lasso, the result will contain also cells that not belong to the cluster?

Thank you in advance,
Adily

[Jinghao-Chen]

Hi, currently, StereoMap does not support selecting cells of a specific cluster via lasso selection. Following your operation method, the result will still include cells that do not belong to the target cluster. We will add support for this feature in the next version. For the time being, we recommend referring to the method provided in this issue to obtain the corresponding cellbin.gef. Once you have obtained the CGEF file for your target cluster, you may then proceed with the Lasso operation on it.
#262

[adilysyn]

Thank you.
Can you please refer to this issue 264 where it was recommended (in the last response) to perform a comparison between two ROI of two experiments, firstly by using the lasso tool. But if the lasso does not support selecting cells of a specific cluster, how can I obtain the comparison between my two ROIs?

[Jinghao-Chen]

1. It is recommended to first export the CGEF files for the specific cluster from both experiments. Refer to the method provided in GEF by cluster #262
2. You may then use StereoMap to perform ROI selection on these two CGEF files output from step 1.
3. Finally, you can run saw reanalyze lasso to generate the corresponding GEF files for comparison.

[adilysyn]

I would like to compare ROIs that composed of several clusters. Should I concatenate CGEFs of different clusters and than upload to StereoMap?

[Jinghao-Chen]

Yes, you may consider using this modified script, which supports inputting multiple clusters separated by commas.

#Please set these parameters
sn=C04042E2 #SN number
cellbin_h5ad=/path/to/outs/analysis/${sn}.cellbin_1.0.adjusted.h5ad
cellbin_gef=/path/to/outs/feature_expression/${sn}.adjusted.cellbin.gef
target_cluster=15,16,17 #the clusters you want to extract
cell_mask=/path/to/outs/image/${sn}_ssDNA_mask_edm_dis_10.tif #If you intend to process the cellbin.gef rather than adjusted.cellbin.gef, please set it to /path/to/outs/image/${sn}_ssDNA_mask.tif
saw_path=/path/to/saw_path #e.g. /path/to/saw-8.2.1
out_path=/path/to/output_directory
raw_gef=/path/to/outs/feature_expression/${sn}.raw.gef

#Then run the attached script get_specific_cluster_cgef_v2.py
${saw_path}/anaconda/bin/python get_specific_cluster_cgef.py ${cellbin_h5ad} ${cellbin_gef} ${target_cluster} ${cell_mask} ${saw_path} ${out_path} ${raw_gef}

The output cellbin.gef file will be saved to ${out_path}.

[adilysyn]

Thank you very much!

[adilysyn]

Thank you, I'll use the updated one.

[adilysyn]

[41][INFO]: start to run filter_cells...
[2026-02-15 06:42:47][Stereo][2748039][MainThread][126494132344640][st_pipeline][44][INFO]: filter_cells end, consume time 0.3392s.
Processing labels:  10%|█▍             | 35636/364094 [57:39<9:28:59,  9.62it/s]

There is a way to make it run faster? only 10% after an hour. I have 64 cpu cores.

[adilysyn]

Another, more critical question: I used the get_specific_cluster_cgef_v2.py script to generate CGEF files for several clusters and uploaded them to StereoMap. However, when selecting the “Cluster” layer, no clusters are visible. When selecting the “Heatmap” layer, only some unrelated clusters appear.

The purpose is to perform ROI selection on the CGEF file, but this is currently not possible.

Could you please help determine why the script output does not match the original clustering (as seen in the source data)?

[Jinghao-Chen]

The script provided earlier only extracts the expression matrix of cells in the corresponding clusters, but does not record the cluster information. Since the current version of StereoMap does not support this function yet, you may provide these files and the clusters you are interested in if convenient., and I will help you with the extraction.

[adilysyn]

Thank you for the clarification.

By “help with the extraction,” do you mean selecting the ROI for the clusters of interest? We are working with several experiments, each containing multiple clusters, and we need to define ROIs for each cluster.

Is there currently any way for us to perform this extraction ourselves? If not, how should we provide you with the relevant files and cluster information?

Thank you for your help,
Adily

[Jinghao-Chen]

@adilysyn Hi, I want to help you extract the cgef files of specific clusters while retaining the cluster information. The results I extracted can be used for your ROI selection in StereoMap. I noticed an error occurred when you ran the script. If you could provide the following files for one of the samples (the sample where the error occurred), it would help me fix the script and assist you with analyzing other samples.

${sn}.cellbin_1.0.adjusted.h5ad
${sn}.adjusted.cellbin.gef
${target_cluster}: e.g. 15,16,17
${sn}_ssDNA_mask_edm_dis_10.tif
${sn}.raw.gef

[adilysyn]

Thank you, I attach here a link to the files: extract_cgef_files

target_cluster: 15_9_6_2_1

[Jinghao-Chen]

@adilysyn Hello, below is the CGEF file obtained after extracting the specific cluster. You may check whether it meets your requirements. I have also updated the script for efficient processing, and it can be directly run with just one step.

https://drive.google.com/file/d/1CCGjLOGSLvTIYd5InAolu9gO0nIBF9Py/view?usp=sharing

get_specific_cluster_cgef_v3.py