DockCert

Automated Statistical Validation, Early Enrichment Metrics, and Reproducibility Assessment for Molecular Docking and Virtual Screening Studies.

Overview

DockCert is an open-source scientific toolkit engineered to evaluate whether a molecular docking or virtual screening protocol is rigorously validated before publication.

Instead of computing enrichment factors or pose RMSD across fragmented scripts, dockcert analyzes your screening results and docked structures in a single run:

• 🎯 Redocking & Cross-Docking Pose Accuracy: Heavy-atom & symmetry-corrected RMSD (Hungarian algorithm).
• 📈 Early Enrichment Quantification:
  • BEDROC ($\alpha=20.0, 80.5, 160.9$)
  • RIE (Robust Initial Enhancement)
  • EF1%, EF5%, EF10% (Enrichment Factors)
  • ROC-AUC & PR-AUC (Precision-Recall AUC)
  • logAUC & Optimal MCC (Matthews Correlation Coefficient)
• 🎲 95% Stratified Bootstrap Confidence Intervals for every metric.
• 🔬 Decoy Bias & Artificial Enrichment Audit: Kolmogorov-Smirnov & Wasserstein property tests (MW, LogP, HBD, HBA).
• 🚦 Quality Certification Badges (PASS / WARNING / FAIL).
• 📑 Publication Deliverables: Interactive self-contained report.html, vector plots (SVG/PDF/PNG 300 DPI), LaTeX summary tables (.tex), and a draft Methods & Supporting Information paragraph with automated BibTeX citations.
• 🔌 Engine Agnostic: Works seamlessly with AutoDock Vina, Smina, GNINA, Schrödinger Glide, CCDC GOLD, DOCK, rDock, and generic CSV tables.

  Docking Results (.csv, .sdf, .pdbqt, logs)
                     │
                     ▼
  ┌───────────────────────────────────────────────────────────┐
  │                         DockCert                          │
  │  ├── Redocking RMSD (Symmetry corrected)                  │
  │  ├── Early Enrichment (BEDROC, RIE, EF1%, ROC-AUC)        │
  │  ├── Stratified 95% Bootstrap Confidence Intervals        │
  │  └── Decoy Bias Audit (MW, LogP KS-tests)                 │
  └───────────────────────────────────────────────────────────┘
                     │
                     ▼
  ┌───────────────────────────────────────────────────────────┐
  │                   Publication Deliverables                │
  │  ├── report.html (Interactive Dashboard & Badges)         │
  │  ├── dockcert_validation_overview.pdf/svg/png             │
  │  ├── dockcert_summary_table.tex / .csv                    │
  │  ├── methods_snippet.txt (Ready for Manuscript)           │
  │  └── citation.bib (BibTeX Reference)                      │
  └───────────────────────────────────────────────────────────┘

Installation

From PyPI

pip install dockcert

From Source

git clone https://github.com/sircalch/dockcert.git
cd dockcert
pip install -e .[dev]

Quickstart (CLI)

1. Run Demonstration Mode (Instant Benchmark Dataset)

dockcert demo -o my_docking_validation/

Open my_docking_validation/report.html in your browser!

2. Assess Virtual Screening CSV

dockcert assess -i screening_results.csv --score-col docking_score --label-col is_active -o validation_report/

3. Assess with Reference Ligand Pose RMSD

dockcert assess -i screening_results.csv --ref-ligand crystal_ligand.sdf --docked-pose docked_pose.sdf -o full_report/

Python API Usage

import numpy as np
from dockcert import assess_docking_quality
from dockcert.reporters import generate_docking_figures, generate_docking_manuscript_assets, generate_docking_html_report

# Screening scores (e.g. 50 actives, 1000 decoys)
labels = np.array([1]*50 + [0]*1000)
scores = np.concatenate([np.random.normal(-9.2, 0.8, 50), np.random.normal(-6.5, 1.1, 1000)])

# Assess quality
report = assess_docking_quality(
    labels=labels,
    scores=scores,
    rmsd_values=[1.42, 1.85, 2.30],
    lower_is_better=True
)

print(f"Overall Validation Status: {report.overall_status}")
print(f"ROC-AUC: {report.enrichment_metrics['roc_auc'].value:.3f}")
print(f"BEDROC (alpha=20.0): {report.enrichment_metrics['bedroc_20'].value:.3f}")
print(f"EF 1%: {report.enrichment_metrics['ef_1pct'].value:.1f}x")

# Export publication deliverables
generate_docking_figures(labels, scores, report, "output_dir/")
generate_docking_manuscript_assets(report, "output_dir/")
generate_docking_html_report(report, "output_dir/report.html")

Citation

If you use DockCert to evaluate molecular docking validation or virtual screening enrichment, please cite:

@software{monreal2026dockcert,
  author = {Monreal-Hern{\'a}ndez, Andre},
  title = {{DockCert: An Open-Source Toolkit for Statistical Validation, Enrichment Metrics, and Reproducibility Assessment of Molecular Docking Studies}},
  year = {2026},
  version = {1.0.0},
  publisher = {Zenodo},
  url = {https://github.com/sircalch/dockcert}
}

License

This project is licensed under the MIT License - see the LICENSE file for details.