snakemake-vizgen-cell-segmentation

A snakemake pipeline to perform cell segmentation on MERFISH spatial transcriptomics data.

Fetaures

This snakemake pipeline is a wrapper around the VizGen Post-processing Tool, to perfrom cell segmentation on MERFISH spatial transcriptomics data.

Build Docker image

• Create image: docker build -t vizgen-segmentation .
• Test image: docker run vizgen-segmentation
• Mount disk: docker run --rm -it --entrypoint /bin/bash -v <path-to-folder>:<docker-folder-name> vizgen-segmentation

Run test on pipeline

• Using watershed segmentation algorithm: bash run_test.sh watershed zero
• Using cellpose model for segmentation:
  • experimental model nuclei-only: bash run_test.sh cellpose three

Run the pipeline with desired VPT model

• Using watershed segmentation algorithm: bash run.sh watershed zero <vgz-file-name>
• Using cellpose model for segmentation:
  • experimental model nuclei-only: bash run.sh cellpose three
  • experimental model cytoplasm with nuclei Z3: bash run.sh cellpose two <vgz-file-name>
  • experimental model cytoplasm with nuclei Z3: bash run.sh cellpose one <vgz-file-name>

Rules input/output

[X] rule identify_cell_boundaries: - input - \images*.tif - \images\micron_to_mosaic_pixel_transform.csv - output - \cellpose_micron_space.parquet - \cellpose_mosaic_space.parquet - \segmentation_specification.json - \results_tiles\0.parquet - \results_tiles\1.parquet - \results_tiles\2.parquet

[X] rule partition_transcripts_cells: - input: - \detected_transcripts.csv - \cellpose_micron_space.parquet - output: - \cell_by_gene.csv - \detected_transcripts.csv

[X] rule calc_cell_metadata: - input: - \cellpose_micron_space.parquet - \cell_by_gene.csv - output: - \cell_metadata.csv

[X] rule calc_cell_sum_signal: - input: - \images*.tif - \images\micron_to_mosaic_pixel_transform.csv - \cellpose_micron_space.parquet - output: - \sum_signal.csv

[X] rule update_vizgen: - input: - \cellpose_micron_space.parquet - \cell_by_gene.csv - \cell_metadata.csv - <project_name>-vizgen-file.vgz - output: - \yy-mm-dd_<project_name>updated.vgz

Structure of output

<path-to-output>\
|   <sample-name>\
|       results_tiles\
|           0.parquet
|           1.parquet
|           2.parquet
|       cellpose_micron_space.parquet
|       cellpose_mosaic_space.parquet
|       segmentaion_segmentation.json
|       cell_by_gene.csv
|       detected_transcripts.csv
|       cell_metadata.csv
|       sum_signals.csv
|       <yy_mm_dd_hh_mm_ss_<project_name>_updated.vgz>