This repo includes all of the code used in the publication by Hoekstra & Slagter et al.: 'Distinct spatiotemporal dynamics of CD8+ T cell-derived cytokines in the tumor microenvironment'. This project involved the use of single-cell transcriptomic read-outs to infer the dissemination of T cell released cytokines through the tumor microenvironment.

This is a view into the code that preprocessed the raw data and generated the figures for use in the manuscript. The raw data is available on GEO. The associated preprocessed data is available on Mendeley data.

All analyses were done in R 4.1. Preprocessing is done using the targets pipelining system, rules are included in the targets directory.

This repo assumes its root directory to be sym-linked to ~/MirjamHoekstra. A set of R variables is asssumed to be encoded in an .Renviron file located under the project root, of which I include here a modified/mock version for your reference.

General reproduction instructions (untested)

• Install a version R.4.1 or up on a Unix-like system
• git clone this repo, e.g. git clone https://github.com/slagtermaarten/cyto_inference.git, on the system from step 1
• (Sym)link the resulting directory to ~/MirjamHoekstra: ln -s ./cyto_inference ~/MirjamHoekstra
• Change directory to cyto_inference/~/MirjamHoekstra, install the project's dependencies by running Rscript bin/install_deps.R. renv and I have never gotten to be friends.
• Update the .Renviron configuration file to reflect your preferences and file system
• Download the preprocessed (single cell) RNASeq data from Mendeley and place the resulting directories in a dir called data. Rename the experiment dirs and add the prefix raw_ to each of them, i.e. cd data; find ./ -name 'exp' -type d -maxdepth 1 -exec mv {} raw_{} \;
• Go back to the project root and run Rscript bin/tar_make to populate the targets object store
• You should now be ready to explore the notebooks and access the targets objects that are referenced within them