How xutaoguo55/cade works
CADE is an R/Python research workflow for ranking how sensitive bulk differential-expression coefficients are to marker-derived cell-composition adjustment, without requiring matched single-cell references. The documented path is bulk expression, group labels and marker sets → relative composition weights → paired limma models → per-gene CCI and stability/calibration summaries. The repository also includes benchmark and validation analyses, result tables, and figure-generation scripts. Edges follow the README’s documented workflow and the sampled figure scripts’ visible table reads; most analysis-script internals were not sampled.
Main entry points: run_all.R Workflow driver · cade_method.R CADE method · raw_csv_components Supplementary tables · cade_external_validation_v4.R External validation
Components 23
CADE workflow
- Workflow driver
code/run_all.R - CADE method
code/cade_method.R - Composition weights
code/cade_method.R - Paired limma models
code/cade_method.R - CCI outputs
code/cade_method.R
Robustness and benchmarks
- ILR sensitivity
code/cade_ilr_uncertainty.R - Multiseed stability
code/validate_cade_ilr_multiseed.R - Method benchmarks
code/benchmark_method_comparison.R - Runtime benchmarks
code/empirical_comparator_runtime_benchmark.R - Model diagnostics
code/generate_model_metadata_table.R
Validation and biology
- External validation
code/cade_external_validation_v4.R - scRNA benchmarks
code/cade_real_scRNA_benchmark_v2.R - PBMC benchmarks
code/cade_pbmc_benchmark.R - WGCNA analysis
code/wgcna_standalone.R
Results and figures
- Result tables
tables - Workflow figure
code/generate_figure1_workflow_summary.py - Benchmark figure
code/generate_figure2_benchmark_clean.py - Pathway figure
code/generate_figure5_pathway_clean.py - WGCNA figure
code/generate_figure6_wgcna_clean.py - Sepsis figure
code/generate_supp_figure_S2_sepsis.py - Supplementary tables
supplementary/raw_csv_components
Other components
- Researcher
- Expression and markers
Connections 22
- Researcher → Workflow driver: runs
README.md - Expression and markers → CADE method: provides inputs
README.md - Workflow driver → CADE method: runs workflow
README.md - CADE method → Composition weights: estimates
README.md - CADE method → Paired limma models: fits
README.md - CADE method → CCI outputs: reports
README.md - Workflow driver → ILR sensitivity: runs analysis
- Workflow driver → Multiseed stability: runs checks
- Workflow driver → Method benchmarks: runs benchmarks
- Workflow driver → Runtime benchmarks: runs benchmarks
- Workflow driver → Model diagnostics: runs diagnostics
- Workflow driver → External validation: runs validation
- Workflow driver → scRNA benchmarks: runs benchmarks
- Workflow driver → PBMC benchmarks: runs benchmarks
- Workflow driver → WGCNA analysis: runs analysis
- CCI outputs → Result tables: summarized in
- Result tables → Benchmark figure: read by
code/generate_figure2_benchmark_clean.py - Supplementary tables → WGCNA figure: read by
code/generate_figure6_wgcna_clean.py - Supplementary tables → Pathway figure: read by
code/generate_figure5_pathway_clean.py - Supplementary tables → Sepsis figure: read by
code/generate_supp_figure_S2_sepsis.py - WGCNA analysis → WGCNA figure: summarized by
- External validation → Sepsis figure: summarized by