How patrickjchoi/antitnf-biomarker-signature works
This repository supports a transcriptomic research workflow: GEO cohort data are harmonized, evaluated for an anti-TNF response signature, checked in blood and an external biopsy cohort, and reported through saved results and manuscript figures. The README documents the main inputs, outputs, and analysis responsibilities; sampled excerpts do not establish every script-to-script dependency, so uncertain links are omitted.
Main entry points: run_blood_ml_optimized.py Blood negative control · generate_all_preprint_figures.py Figure generation · run_batch_corrected_loocv.py Batch-corrected CV · gse73661_analysis.py GSE73661 validation
Components 24
Cohort preparation
- Biopsy harmonization
Code/harmonize_biopsy.py - Blood harmonization
Code/harmonize_blood.py - Biopsy metadata
harmonized_biopsy_metadata.pkl - Blood metadata
harmonized_blood_metadata.pkl
Signature discovery
- Batch-corrected CV
Code/run_batch_corrected_loocv.py - Batch CV results
batch_corrected_loocv_results.pkl - Batch permutation test
Code/run_batch_corrected_permutation.py - Batch permutation results
batch_corrected_permutation_results.pkl - Leave-one-dataset-out
Code/run_batch_corrected_lodo.py - LODO results
batch_corrected_lodo_pooled_results.pkl - Feature ranking
Code/analyze_batch_corrected_features.py - Feature rank results
feature_rank_table_results.pkl - Subgroup evaluation
Code/run_subgroup_analysis_v2.py - Subgroup results
subgroup_analysis_v2_results.pkl
Validation and controls
- Blood negative control
Code/run_blood_ml_optimized.py - Blood ML results
blood_ml_full_results.pkl - Blood permutation results
blood_permutation_results.pkl - GSE73661 validation
Code/gse73661_analysis.py - Validation results
gse73661_validation_results.pkl
Research outputs
- Figure generation
Code/generate_all_preprint_figures.py - Generated figures
Results - Research manuscript
Choi_Harmonized_AntiTNF_Biomarker_Manuscript.md
Other components
- Researcher
- NCBI GEO cohorts
Connections 29
- Researcher → Biopsy harmonization: runs
- Researcher → Blood harmonization: runs
- NCBI GEO cohorts → Biopsy harmonization: provides cohorts
- NCBI GEO cohorts → Blood harmonization: provides cohorts
- Biopsy harmonization → Biopsy metadata: writes metadata
- Blood harmonization → Blood metadata: writes metadata
- Researcher → Batch-corrected CV: runs
- Researcher → Batch permutation test: runs
- Researcher → Leave-one-dataset-out: runs
- Researcher → Feature ranking: runs
- Researcher → Subgroup evaluation: runs
- Researcher → Blood negative control: runs
- Researcher → GSE73661 validation: runs
- Batch-corrected CV → Biopsy metadata: reads metadata
- Batch-corrected CV → Batch CV results: writes results
- Batch permutation test → Batch permutation results: writes results
- Leave-one-dataset-out → LODO results: writes results
- Feature ranking → Feature rank results: writes results
- Subgroup evaluation → Subgroup results: writes results
- Blood negative control → Blood metadata: reads metadata
- Blood negative control → Blood ML results: writes results
- Blood negative control → Blood permutation results: writes results
- GSE73661 validation → NCBI GEO cohorts: retrieves cohort
- GSE73661 validation → Validation results: writes results
- Researcher → Figure generation: runs
- Figure generation → Batch CV results: reads metrics
- Figure generation → Validation results: reads results
- Figure generation → Generated figures: writes figures
- Researcher → Research manuscript: consults