End-to-End Tissue Microarray Image Analysis with Galaxy-ME

Use this workflow when you have raw, unprocessed cycle images that need full preprocessing before analysis.

Input datasets

• Collection of raw cycle images (TIFF/OME-TIFF): Ensure that the list is ordered in cycle order (ex: cycle_1.tiff, cycle_2.tiff, etc.)

• Markers file (CSV): A comma-separated file with marker_names in the third column

  • Example markers file:

round,channel,marker_name
0,0,DAPI_1
0,1,CD3
0,2,CD45
0,3,CD8
1,4,DAPI_2
1,5,PANCK
1,6,SMA
1,7,ECAD
...

• Phenotype file (CSV): A comma-separated Scimap phenotyping file that maps hierarchical cell phenotypes to markers

  • For an example phenotype workflow, see our tutorial and the Scimap documentation.

Input values

All input values have been preset in the workflow and are optimized for cyclic immunofluorescence images captured using a Rarecyte slide scanner. Some important assumptions are made:

• Channel used as a reference for registration (ASHLAR): 0
• Channel used for nuclear segmentation (Mesmer): 0
• Image resolution (microns per pixel): 0.65

The workflow should be imported and edited if these values are not suitable for your datasets.

Processing

For more detailed information, see our tutorial on the Galaxy Training Network

• Tile-to-tile illumination differences are corrected in the unstitched input raw cycle images using Basic Illumination
• A whole-slide OME-TIFF image is generated via stitching and registration with ASHLAR. Channel names are assigned at this step using the input markers file
• TMA cores are segmented and cropped into individual images, producing a collection of TIFFs using UNetCoreograph. All subsequent steps are run as batch processing across the collection of core datasets
• The output of UNetCoreograph is a generic TIFF, and must be converted back to OME-TIFF using the Convert Image tool, and channels can be renamed using the Rename OME-TIFF channels utility
• Nuclear segmentation is performed using Mesmer, producing a nuclear mask in TIFF format for each core image
• Cell/nuclear features (mean marker intensities, spatial coordinates, and morphological measurements) are quantified using MCQUANT, producing a CSV table of cells (rows) x features (columns)
• The quantification table is converted to anndata format (h5ad), a common datatype used by most single-cell and spatial analysis packages
• Automated cell phenotyping is performed using Scimap (see Warning section about GMM-based phenotyping)
• Finally, Vitessce dashboards combine interactive image viewing with linked single-cell analysis components to allow for integrated initial data exploration

Warning

In this workflow, we perform automated GMM-based cell phenotyping using Scimap. The Scimap tool also accepts manual gates, which can be determined using the GateFinder tool. This method is highly recommended, as most markers are not well suited for GMM-based thresholding. The automated GMM-based thresholding can work well for highly abundant markers that show a strong bimodal distribution; otherwise, it should be used primarily as a means of generating an initial starting point for gating and cell phenotyping.

For more warnings and context, see our tutorial linked above.

Tool developers' documentation

• MCMICRO
  • Basic Illumination
  • ASHLAR
  • UNetCoreograph
  • MCQuant
• Mesmer
• Scimap
• Vitessce

Tool references

• Peng, T., K. Thorn, T. Schroeder, L. Wang, F. J. Theis et al., 2017 A BaSiC tool for background and shading correction of optical microscopy images. Nature Communications 8: 10.1038/ncomms14836
• Wolf, F. A., P. Angerer, and F. J. Theis, 2018 SCANPY: large-scale single-cell gene expression data analysis. Genome Biology 19: 10.1186/s13059-017-1382-0
• Stringer, C., T. Wang, M. Michaelos, and M. Pachitariu, 2020 Cellpose: a generalist algorithm for cellular segmentation. Nature Methods 18: 100–106. 10.1038/s41592-020-01018-x
• Greenwald, N. F., G. Miller, E. Moen, A. Kong, A. Kagel et al., 2021 Whole-cell segmentation of tissue images with human-level performance using large-scale data annotation and deep learning. Nature Biotechnology 40: 555–565. 10.1038/s41587-021-01094-0
• Schapiro, D., A. Sokolov, C. Yapp, Y.-A. Chen, J. L. Muhlich et al., 2021 MCMICRO: a scalable, modular image-processing pipeline for multiplexed tissue imaging. Nature Methods 19: 311–315. 10.1038/s41592-021-01308-y Virshup, I., S. Rybakov, F. J. Theis, P. Angerer, and F. A. Wolf, 2021 anndata: Annotated data. 10.1101/2021.12.16.473007
• Muhlich, J. L., Y.-A. Chen, C. Yapp, D. Russell, S. Santagata et al., 2022 Stitching and registering highly multiplexed whole-slide images of tissues and tumors using ASHLAR (A. Valencia, Ed.). Bioinformatics 38: 4613–4621. 10.1093/bioinformatics/btac544
• Palla, G., H. Spitzer, M. Klein, D. Fischer, A. C. Schaar et al., 2022 Squidpy: a scalable framework for spatial omics analysis. Nature Methods 19: 171–178. 10.1038/s41592-021-01358-2
• Yapp, C., E. Novikov, W.-D. Jang, T. Vallius, Y.-A. Chen et al., 2022 UnMICST: Deep learning with real augmentation for robust segmentation of highly multiplexed images of human tissues. Communications Biology 5: 10.1038/s42003-022-04076-3
• Zhang, W., I. Li, N. E. Reticker-Flynn, Z. Good, S. Chang et al., 2022 Identification of cell types in multiplexed in situ images by combining protein expression and spatial information using CELESTA. Nature Methods 19: 759–769. 10.1038/s41592-022-01498-z
• Nirmal, A. J., and P. K. Sorger, 2024 SCIMAP: A Python Toolkit for Integrated Spatial Analysis of Multiplexed Imaging Data. Journal of Open Source Software 9: 6604. 10.21105/joss.06604