jazzPanda: Paper Repository

This repository provides the complete set of R scripts used to generate all main and supplementary figures for the jazzPanda paper.

• jazzPanda R package: Repository
• Preprint: bioRxiv link
• Raw/Processed data: Download from Zenodo
• Analysis workflow: Workflow documentation

.
├── scripts/
│   ├── main/
│   └── supp/
├── figures/                  # outputs (PDF/JPG) — created by scripts
│   ├── main/
│   └── supp/
├── data/
├── .gitignore
└── README.md                 # you are here

Datasets

Four imaging-based spatial transcriptomics datasets are used throughout the paper:

Short name	Platform	Tissue
xenium_hbreast	10x Xenium	Human HER2+ breast cancer (2 samples)
cosmx_hhliver	Nanostring CosMx	Human healthy liver
cosmx_hlc	Nanostring CosMx	Human liver cancer
merscope_hbreast	Vizgen MERSCOPE	Human breast cancer

Data Availability

Raw experimental data can be accessed via the links below. Processed outputs used in the manuscript are included in the repository under the data/ directory, with intermediate results stored as .Rds files. For very large processed datasets that exceed the repository size limits, we provide access via Zenodo.

Raw data

-1) 10x Xenium Human HER2+ breast cancer data 🔗

-2) Nanostring CosMx Human liver healthy and cancer data 🔗

-3) Vizgen MERSCOPE Human breast cancer data 🔗

Processed data

Processed outputs are organized in the data/ directory of this repository, with intermediate results saved as .Rds files. Very large processed datasets that exceed repository size limits are deposited on Zenodo for convenient download.

Figure map

Figure	Content	Script	Output
Figure 1	Data sparsity and performance of single-cell marker analysis methods	scripts/main/figure_gene_count.R	figures/main/figure_intro/
Figure 2	Method overview (schematic)	—	—
Figure 3	Negative control probes and simulation-based evaluation	scripts/main/figure_simulation.R	figures/main/figure_simulation/
Figure 4	Application to CosMx healthy liver and Xenium HER2+ breast cancer	scripts/main/figure_result.R	figures/main/figure_result/
Figure 5	Comparison of marker analysis methods	scripts/main/figure_compare_methods.R	figures/main/figure_compare_methods/
Figure 6	Extension of the jazzPanda framework (MERSCOPE breast cancer)	scripts/main/figure_sv_extension.R	figures/main/figure_sv_extension/
Figure 7	Technical performance (bin size, runtime, memory)	scripts/main/figure_technical_performance.R	figures/main/figure_technical_performance/

Script Overview

Marker gene detection

Location	Scripts	Output
scripts/main	run_mg_xenium_human_breast_cancer.sh mg_xenium_human_breast_cancer.R	data/dataset_computational_complexity/xenium_hbreast_*.Rds
scripts/main	run_mg_cosmx_human_healthy_liver.sh mg_cosmx_human_healthy_liver.R	data/dataset_computational_complexity/cosmx_hhliver_*.Rds
scripts/main	run_mg_cosmx_human_liver_cancer.sh mg_cosmx_human_liver_cancer.R	data/dataset_computational_complexity/cosmx_hlc_*.Rds
scripts/main	run_mg_merscope_human_breast_cancer.sh mg_merscope_human_breast_cancer.R	data/dataset_computational_complexity/merscope_hbreast_*.Rds

Simulation

Analysis	Location	Scripts	Output
Simulation: CosMx human liver cancer	scripts/main	cosmx_hlc_simulation_simbg_sa.sh cosmx_hlc_simulation_simbg_slurmarray_temp.R	scripts/main/cosmx_hlc_simulation_result/

Technical performance

Tiles on marker genes

Location	Scripts	Output
scripts/main/discussion_markergenes_vs_ntiles	xenium_hbreast_ntile_markergenes_sa.sh xenium_hbreast_ntile_markergene_result.R	scripts/main/discussion_markergenes_vs_ntiles/xenium_hbreast_ntiles_mg_gr{10–100}_*.csv

Computational complexity on number of cores

Location	Scripts	Output
scripts/main/discussion_complexity	complexity_ncores_sa.sh complexity_ncores_cosmx_hliver_cancer_slurmarray_temp.R	scripts/main/discussion_complexity/ncores_result/

Computational complexity on number of transcripts

Location	Scripts	Output
scripts/main/discussion_complexity	complexity_ntr_sim_sa.sh complexity_ntr_simulation.R	scripts/main/discussion_complexity/ngenes_result/

Computational complexity on number of tiles

Location	Scripts	Output
scripts/main/discussion_complexity	complexity_ntiles_squarebins_sim_sa.sh complexity_ntiles_squarebin_simulation.R	scripts/main/discussion_complexity/ntiles_squarebins_result/
scripts/main/discussion_complexity	complexity_ntiles_hexbin_sim_sa.sh complexity_ntiles_hexbin_simulation.R	scripts/main/discussion_complexity/ntiles_hexbin_result/

Supplementary material

Supplementary figures and tables are generated as R Markdown notebooks in scripts/supp/, each rendered to a self-contained .html of the same name.

Notebook	Content
supplementary_simulation.Rmd	Permutation p-values and correlation of simulated genes across clusters; false discovery rates
supplementary_application_cosmx_healthy_human_liver.Rmd	Dataset overview; top marker genes per cluster and cluster–gene vector relationships (jazzPanda-correlation and jazzPanda-glm); marker gene overlap and cumulative average correlation across methods
supplementary_application_cosmx_human_liver_cancer.Rmd	Dataset overview
supplementary_application_xenium_human_breast_cancer.Rmd	Dataset overview; top marker genes per cluster and cluster–gene vector relationships (jazzPanda-glm); marker gene overlap and cumulative average correlation across methods; marker gene overlap across tile lengths
supplementary_application_merscope_human_breast_cancer.Rmd	Dataset overview; spatial visualisation of selected cell types
supplementary_technical_performance.Rmd	Computational complexity of spatial vector construction across square and hex bin lengths; runtime and memory benchmarks