♻️ Coralysis-reproducibility

Code to reproduce the analyses from the Coralysis manuscript:

António GG Sousa, Johannes Smolander, Sini Junttila, Laura L Elo (2025).
Coralysis enables sensitive identification of imbalanced cell types and states in single-cell data via multi-level integration. bioRxiv. https://doi.org/10.1101/2025.02.07.637023

🗂️ Scripts & Datasets

All the scripts required to reproduce the Coralysis manuscript figures described in the table below are available in this repository under the folder scripts, with the exception of the code from the last row, which has its own repository and respective documentation at elolab/scib-pipeline.

All the datasets used in the manuscript are publicly available online, and the links are provided in the table below. A more detailed description of the datasets and their respective references can be found at data/code_datasets_description.tsv.

🔗 Summary of scripts used to reproduce the Coralysis manuscript figures.

Scripts	Description	Data
01_figure_1.R	R script to make Figure 1	Assay V1; Assay V2
02_extdata_figure_2.R	R script to make Supplementary Figure S1	Assay V1; Assay V2
03_figure_2.R	R script to make Figure 3B & Supplementary Figure S12	ifnb from SeuratData (parsed)
04_extdata_figure_3_4_5.R	R script to make Figure 2 & Supplementary Figure S8-S11	Figshare
05_figure_3.R	R script to to make Figure 5A-G,I	GitHub
06_figure_3.py	Python script to make Figure 5H,M & Supplementary Figure S15	GitHub
07_figure_4.R	R script to make Figure 6	panc8 from SeuratData
08_extdata_figure_6.R	R script to make Supplementary Figure S16,S17	Figshare
09_extdata_figure_7.R	R script to make Supplementary Figure S18,S19	Figshare
10_extdata_figure_8.R	R script to make Supplementary Figure S20	ifnb from SeuratData
11_figure_5.R	R script to make Figure 7	pbmcsca from SeuratData
12_figure_6.R	R script to make Figure 8 & Supplementary Figure S23	Zenodo; Figshare
13_mapref_switching_refquery.R	R script to make Supplementary Figure S21,S22	panc8 from SeuratData; ifnb from SeuratData; Figshare
14_benchmark_imbalance.R	R script to make Figure 4B & Supplementary Figure S14	Figshare
helper_functions.R	R script with custom R functions
elolab/scib-pipeline	Snakemake workflow to benchmark Coralysis - Figure 3C,4A & Supplementary Figure S2-S7,S13	Figshare; ifnb from SeuratData (parsed); Figshare (parsed)

🛠️ Reproducibility

All R scripts were run with R version 4.2.1 under the RStudio Server environment (v.2022.07.2 Build 576), deployed through the Docker image elolab/sctoolkit.

The analyses can be reproduced using the Coralysis version corresponding to the commit 47f1b3415663ee895df188f264ac4d8ad8d24c11, which can be installed as follows:

devtools::install_github("elolab/Coralysis", ref = "47f1b3415663ee895df188f264ac4d8ad8d24c11")

The remaining R package dependencies and their respective versions can be found at the end of every R script (obtained with the R command sessionInfo()). The R scripts were run in the order specified by the prefix in each file name — i.e., 01_figure_1.R, followed by 02_extdata_figure_2.R, and so on.

The Python script (06_figure_3.py) was run with Python version 3.9.16, along with the packages numpy (v.1.26.4), scanpy (v.1.10.3), and scib_metrics (v.0.5.1).

The elolab/scib-pipeline benchmark was run with Snakemake (v.7.25.2) in a cluster environment using Slurm (v.23.02.6) with 8 or 100 threads and 354 or 384 GB of RAM. The respective conda environments can be found under envs (scib-R4.1.yml and scib-pipeline-R4.1.yml). The configuration files used for the benchmark are available at:

• benchmark_coralysis.yaml

• benchmark_imbalance.yaml

After activating the Snakemake conda environment, the benchmark was initiated with the following commands:

# Main benchmark across 6 datasets plus ifnb dataset comprising unshared similiar cell type pairs
snakemake --configfile configs/benchmark_coralysis.yaml --cores 8 

# Benchmark imbalanced cell type integration
snakemake --configfile configs/benchmark_imbalance.yaml --cores 30 

The integration performance was summarised and visualised using custom R functions used in Luecken et al. (2022) and available at: https://github.com/theislab/scib-reproducibility/tree/main/visualization.

🏛️ Funding

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement no.: 955321.