Skip to content

use_case.md

Latest commit

 

History

History
157 lines (104 loc) · 2.8 KB

use_case.md

File metadata and controls

157 lines (104 loc) · 2.8 KB
Raw

How to use GLOOM

A. Run the full pipeline

Use this when you want everything from raw data to final results.

cd script
python run_pipeline.py

Best for

  • first full analysis
  • final production run
  • reproducible end-to-end results

B. Run step by step

Use this when you want control or debugging.

python config.py
python step1_data_loading.py
python step2_preprocessing.py
python step3_harmonization.py
python step4_differential_expression.py
python step5_expression_features.py
python step6_coexpression_network.py
python step7_network_features.py
python step8_feature_integration.py
python step9_label_construction.py
python step10_train_val_split.py
python step11_model_training.py
python step12_model_evaluation.py
python step13_feature_importance.py
python step14_gene_ranking.py
python step15_network_annotation.py
python step16_network_export.py
python step17_interactive_visualization.py
python step18_final_report.py

Best for

  • debugging
  • changing one parameter and rerunning only later steps
  • teaching/demo use

C. Use GLOOM only for gene ranking

If your real question is: “Which genes should I study first?”

Use:

  • step14_gene_ranking.py
  • results/gene_rankings.csv
  • results/novel_candidates.csv

Best for

  • candidate gene discovery
  • shortlist generation
  • downstream manual review

D. Use GLOOM only for model comparison

If your real question is: “Which machine-learning model worked best?”

Use:

  • step12_model_evaluation.py
  • results/model_metrics.csv
  • ROC / PR figures

Best for

  • ML benchmarking
  • method comparison
  • choosing the model for deployment

E. Use GLOOM only for interpretation

If your question is: “Why did the model choose these genes?”

Use:

  • step13_feature_importance.py
  • results/feature_importance.csv
  • grouped importance figure

Best for

  • explainability
  • reports and manuscripts
  • understanding model behavior

F. Use GLOOM only for network exploration

If your question is: “How are these genes connected?”

Use:

  • step15_network_annotation.py
  • step16_network_export.py
  • results/network/annotated_network.graphml
  • results/network/exports/

Best for

  • Cytoscape/Gephi work
  • hub gene analysis
  • subnetwork discovery

G. Use GLOOM only for interactive browsing

If you want an easy visual interface without touching Python tables.

Open:

  • figures/interactive_dashboard.html

Best for

  • collaborators
  • presentations
  • fast exploration

H. Use GLOOM only for final reporting

If you want a compact end product.

Use:

  • results/reports/pipeline_report.txt
  • results/reports/pipeline_summary_table.csv
  • results/reports/pipeline_summary_figure.png

Best for

  • posters
  • slides
  • reporting to supervisors or collaborators