Cardiomyocyte Pseudobulk and Pathway Scoring

Reproducible scripts for analyzing public human myocardial infarction single-nucleus data. The workflow keeps only the core analysis: extraction of Cardiomyocyte raw RNA counts, count-based pseudobulk differential expression, and pathway scoring with curated mitochondrial and inflammatory gene sets.

Analysis Overview

The formal differential-expression analysis uses raw Cardiomyocyte RNA counts from the public Seurat RDS. Counts are aggregated by sample and priority region, normalized with edgeR TMM, and modeled with limma-voom. The primary contrast is IZ_vs_Control.

Mixed anatomical regions are assigned by priority:

Control > FZ > IZ > BZ > RZ

Thus IZ_vs_Control should be interpreted as IZ-associated Cardiomyocyte change, not a pure infarct-zone-only comparison.

Pathway scoring is performed on Control and IZ-associated cells from the processed H5AD using UCell-like and AUCell-like rank-based scores.

Primary contrast sample inclusion:

Group	Sample	Patient	Source region label	Cardiomyocyte RNA nuclei
Control	CK158	P1	control_P1	2,637
Control	CK357	P7	control_P7	4,460
Control	CK358	P8	control_P8	2,715
Control	CK374	P17	control_P17	3,173
IZ-associated	CK165	P2	IZ/BZ_P2	601
IZ-associated	CK361	P10	IZ_P10	705
IZ-associated	CK363	P9	GT/IZ_P9	810
IZ-associated	CK371	P15	GT/IZ_P15	305
IZ-associated	CK372	P9	GT/IZ_P9_rep2	159

Inputs

Public data files are downloaded by the workflow and are not committed to GitHub:

results/raw/snRNA-seq-submission.h5ad
results/raw/Cardiomyocyte_snRNA_snATAC.Rds

The gene-set preparation step expects these source files under gene_sets_official/raw/:

h.all.v2025.1.Hs.symbols.gmt
c2.cp.reactome.v2025.1.Hs.symbols.gmt
c2.cp.kegg_legacy.v2025.1.Hs.symbols.gmt
c5.go.bp.v2025.1.Hs.symbols.gmt
human.mitocarta3.0.html
Human.MitoPathways3.0.gmx

MSigDB GMT files must be obtained by users under the MSigDB terms of use. MitoCarta and MitoPathways files are available from the Broad Institute MitoCarta resource.

Environment

The environment is managed with pixi:

pixi install

pixi.lock is included for reproducibility on Linux. R scripts run in the r environment; Python scripts run in the analysis environment.

Workflow

Run all commands from the repository root:

pixi run download-data
pixi run -e r extract-counts
pixi run -e r pseudobulk
pixi run -e analysis prepare-gene-sets
pixi run -e analysis score-pathways

Main Outputs

Cardiomyocyte RNA counts:

results/processed/Cardiomyocyte_RNA_counts_only.seurat.rds
results/processed/Cardiomyocyte_RNA_counts_only.metadata.tsv.gz
results/processed/Cardiomyocyte_RNA_counts_only.summary.tsv

Pseudobulk differential expression:

results/pseudobulk_cardiomyocyte/tables/pseudobulk_sample_metadata.tsv
results/pseudobulk_cardiomyocyte/tables/cardiomyocyte_pseudobulk_counts_matrix.rds
results/pseudobulk_cardiomyocyte/tables/edgeR_voom_Cardiomyocyte_IZ_vs_Control.tsv
results/pseudobulk_cardiomyocyte/tables/deg_counts_by_group.tsv
results/pseudobulk_cardiomyocyte/figures/volcano_Cardiomyocyte_IZ_vs_Control.png
results/pseudobulk_cardiomyocyte/figures/heatmap_sig_degs_Cardiomyocyte_IZ_vs_Control.png

Curated gene sets:

gene_sets_official/mitochondria/*.txt
gene_sets_official/inflammation/*.txt
gene_sets_official/metadata/manifest.tsv

Pathway scores:

results/pathway_scores_control_iz/tables/pathway_scores_per_cell_control_iz.tsv.gz
results/pathway_scores_control_iz/tables/pathway_scores_by_sample_celltype_control_iz.tsv
results/pathway_scores_control_iz/tables/pathway_ucell_cardiomyocyte_matrix.tsv
results/pathway_scores_control_iz/tables/pathway_aucell_cardiomyocyte_matrix.tsv
results/pathway_scores_control_iz/figures/heatmap_pathway_ucell_cardiomyocyte.png
results/pathway_scores_control_iz/figures/heatmap_pathway_aucell_cardiomyocyte.png
results/pathway_scores_control_iz/figures/violin_top_cardiomyocyte_ucell_control_iz.png
results/pathway_scores_control_iz/figures/violin_top_cardiomyocyte_aucell_control_iz.png

Interpretation

The pseudobulk DEG analysis is the primary statistical layer because it uses raw counts and sample-level replication. Pathway scores provide supporting cell-level evidence for coordinated biological programs.

A cautious summary is:

IZ-associated Cardiomyocytes show reduced mitochondrial oxidative phosphorylation programs and increased inflammatory or cytokine-related programs compared with Control Cardiomyocytes.

Repository Contents

00_download_public_data.sh
01_extract_cardiomyocyte_counts.R
02_pseudobulk_edgeR_voom.R
03_prepare_official_gene_sets.py
04_score_pathways_ucell_aucell.py
pixi.toml
pixi.lock
CITATION.md
LICENSE
README.md

Public data, generated results, and restricted gene-set source files are excluded by .gitignore.