Authors: Clarence Yapp, Ajit J. Nirmal, Felix Zhou, Alex Y.H. Wong, Juliann B. Tefft, Yi Daniel Lu, Zhiguo Shang, Zoltan Maliga, Paula Montero Llopis, George F Murphy, Christine G Lian, Gaudenz Danuser, Sandro Santagata, and Peter K. Sorger

This respository (labsyspharm/mel-3d-mis DOI: 10.5281/zenodo.10055593) hosts original code associated with the above publication for 3D image registration, intensity quantification, cell shape & orientation analysis, collagen-to-cell distance quantification, and cell interaction analysis. See Data Access section (below) for instructions on accessing the primary data associated with this paper.

Please cite this data as the following: Yapp, C., Nirmal A.J. et al. (2025). Highly Multiplexed 3D Profiling of Cell States and Immune Niches in Human Tumours, Nature Methods.

Relevant links:

• Publication DOI: https://www.biorxiv.org/content/10.1101/2023.11.10.566670v4.
• Associated GitHub Repository: labsyspharm/mel-3d-mis
• To view an archived record of this repository: https://zenodo.org/records/15230302
• To view the image data online, visit: (https://www.tissue-atlas.org/atlas-datasets/yapp-nirmal-2023)
• DOI of other publications that use the data:
  • Multimodal Spatial Profiling Reveals Immune Suppression and Microenvironment Remodeling in Fallopian Tube Precursors to High-Grade Serous Ovarian Carcinoma https://doi.org/10.1158/2159-8290.CD-24-1366
  • Universal consensus 3D segmentation of cells from 2D segmented stacks https://doi.org/10.1101/2024.05.03.592249
  • A Mixed Reality and 2D Display Hybrid Approach for Visual Analysis of 3D Tissue Maps. https://doi.org/10.31219/osf.io/zka2j

​

Licenses/restrictions placed on the data: CC-BY creativecommons.org/licenses/by/4.0/

This demo illustrates how to assess cell type interactions between neighboring cells by measuring membrane intensities along horizontal line intensity profiles. A CD4 and CD8 T cell are used as an example here and is illustrated in Figure 6p-r of corresponding manuscript.

Download code and source image (Figure 6.tif) found in demo folder. The .tif file is a truncated version of Figure 6p from manuscript but with only 4 z planes and 2 channels (CD4 and CD8 respectively). Load the the CD4 and CD8 channels by running the first 2 lines. Choose either tight or loose interaction, and run the corresponding code block. Run remainder of script.

• Cell type interaction analysis, intensity quantification, and cell shape & orientation analysis require a standard computer with 8GB RAM.
• 3D image registration and collagen-to-cell distance quantification require significant amounts of RAM (tested on 600-700GB of RAM on a high performance cluster) and depends on the size of the image ROI. Updated versions will be less RAM intensive and may also feature parallelization over tiles.

MATLAB version 2021/2022 (or newer) on Windows 10 with the following toolboxes:

• Curve Fitting Toolbox (tested on version 3.6)
• Statistics and Machine Learning Toolbox (tested on version 12.2)
• Image Processing Toolbox (tested on version 11.4)
• Parallel Computing Toolbox (OPTIONAL tested on version 7.5) Install time is typically 1 hour or less but will vary depending on network speed and computer specifications.

When complete, the demo will output a graph with two intensity profiles and their fitted polynomial curves. The x-axis is in microns along intensity profile(s). The y-axis is the normalized intensity where 1 is the maximum intensity along the line profile. Red 'x' denote the peak of each polynomial curve and estimates the location of the membrane at sub-pixel accuracy. The separation between both peaks (denoted by red 'x') along x-axis is a measure of how 'tight' the membrane interaction is. For example, the distance between cell 2 and 3 is ~30nm and represents a tight interaction (type I).

The distance between cell 4 and 5 is ~150nm. This is a type II interaction.

Expected run time will be on the order of seconds especially if data is being loaded from local disk. If data is on network, run time will be longer.

Image can be a multichannel 3D file or separated channels. Images should be generated with considerations outlined in Supplementary section of corresponding manuscript (ie. high signal-to-noise ratio, validated antibodies, etc).

1. update file paths to locations of channel A and B (line 1 and 2).
2. images may need to be rotated in order to draw horizontal line intensity profiles. Change variable rot_angle to an angle that rotates the image so that your line profile will horizontally through the cell membrane of interest.
3. enter XY coordinates of the start of line profile (xstart, ystart) and the end (xend, yend). Select optimal Z plane index, planeZ.
4. update pixelSize to reflect the lateral pixel sizes (microns/pixel) in your image data.
5. (optional), change colors of curves following RGB convention on line 38.

​

​ Stitched and registered 3D multiplexed data is available for download through AWS.
​ You will need the following bucket name:

s3://lsp-public-data/yapp-2023-3d-melanoma/

​ For general instructions on how to download data from AWS, see: https://zenodo.org/records/10223574

If you experience issues accessing the above AWS S3 buckets, email tissue-atlas(at)hms.harvard.edu with the subject line "s3://lsp-public-data/yapp-2023-3d-melanoma/: Data Access".
​

​