inst/scripts/README.md

In BodenmillerGroup/SingleCellMapper: Visualization of highly multiplexed imaging data in R

Example data

This folder describes the files contained in inst/extdata, how these files were obtained and how to generate them by running the scripts contained in insts/scripts.

Raw Imaging Mass Cytometry (IMC) data are computationally processed using a segmentation pipeline (available at https://github.com/BodenmillerGroup/ImcSegmentationPipeline). This produces image stacks containing the raw pixel values for > 35 channels, masks containing the segmented cells, cell-level expression, metadata information as well as a number of image-level meta information.

The dataset used here is associated with the publication A Map of Human Type 1 Diabetes Progression by Imaging Mass Cytometry and is available from Mendeley Data: http://dx.doi.org/10.17632/cydmwsfztj.2.

The scripts contained in insts/scripts can be run to download the data and to generate the toy dataset used to illustrate the cytomapper package.

Content of the inst/extdata folder:

This folder contains the tiff files from which the toy dataset in data was generated. There are three images stacks (suffix _imc.tiff), each containing 5 channels for a size of 100 x 100 pixels and three associated cell masks (suffix _mask.tiff). The image names (E34, G01, J02) are used to match the image stacks with the associated mask.

Downloading the dataset

The dataset can be downloaded and saved using the 1_LoadPancreasData.Rmd and the 2_LoadPancreasImages.Rmd scripts.

The 1_LoadPancreasData.Rmd script generates a SingleCellExperiment object: - pancreas_sce.rds: contains the single cell data.

The 2_LoadPancreasImages.Rmd generates two CytoImageList objects: - pancreas_images.rds: contains the multiplexed images. - pancreas_masks.rds: contains the cell masks.

These files are saved in the inst/exdata subfolder and form a dataset comprising a hundred multiplexed images that each contain 38 channels, as well as the associated single-cell data. This represents a subset of the original publication data set, which contains 845 images. The full dataset is also available from Mendeley Data.

Generating the toy dataset

For space reasons, the 100 images dataset needs to be subsetted in order to generate the small toy dataset used to illustrate the cytomapper package.

This toy dataset is generated from the three .rds files generated above by running the 3_GenerateToyData.Rmd script.

In this script, a subset of three images (named E34, G01, and J02) is extracted and these images are cropped to 100 x 100 pixels. The number of channels is limited to five (H3, CD99, PIN, CD8a and CDH).

The SingleCellExperiment object containing the single-cell data is subsetted accordingly in order to retain only the cells and channels that are present on the toy images.

When running the 3_GenerateToyData.Rmd script, the following files are created: In inst/extdata: - imc.tiff files: stacks corresponding to the subsetted images. - mask.tiff files: masks associated with the imc.tiff files.

In data, the toy dataset: - pancreasImages.RData: a CytoImageList object containing the images. - pancreasMasks.RData: a CytoImageList object containing the masks. - pancreasSCE.RData: a SingleCellExperiment object containing the single- cell data.

BodenmillerGroup/SingleCellMapper documentation built on July 2, 2024, 4:31 p.m.