BN-PAH CO₂ Pipeline v0.38

Full pipeline from molecular SMILES → COMPAS-4 descriptors → transport → CO₂ binding → routing → product yield prediction.

---

What this does

SMILES → nDR, nSP, nLL, B/N position → Transport (highcycle_bet_share_ge2)
         → E_ads (kcal/mol) → Binding mechanism (diatomic lens)
         → Routing (nSP parity → 3 paths)
         → Yield (CH3OH / CH4 / HCOOH)
         → Efficiency (capture_eff × routing_eff)

Processes 23,856 BN-PAH molecules from COMPAS-4 database.
Validated against 7 diatomic references (H₂, LiH, F₂, BN, OH, N₂, C₂).
Connected to G11/G12 transport constraints and Fe/NbSe₂ chain parity.

---

Quick start

# Clone
git clone https://github.com/YOUR_USERNAME/bnpah_co2_pipeline.git
cd bnpah_co2_pipeline

# Install dependencies
pip install rdkit networkx scikit-learn pandas numpy matplotlib scipy

# Run analysis
python scripts/bnp_compas_analysis.py
python scripts/co2_pipeline_final.py
python scripts/extended_validation_dft.py

---

Repository structure

bnpah_co2_pipeline/
├── src/
│   ├── bnp_compas_analysis.py       # COMPAS-4 descriptor computation
│   ├── co2_pipeline_final.py        # Full CO2 pipeline (binding→transport→routing)
│   ├── extended_validation_dft.py   # DFT input generation + 7-ring extrapolation
│   ├── generate_ring7_bnpah.py      # 7-ring BN-PAH isomer generation
│   └── final_integrated_pipeline.py # Integrated analysis (all 18 panels)
├── data/
│   ├── compas_results.csv           # Compact results (17 cols, 23,856 molecules)
│   ├── compas_results_full.csv      # Full results + transport features
│   ├── co2_pipeline_results.csv     # CO2 pipeline (26 cols, routing + yields)
│   ├── co2_pipeline_with_g11sectors.csv  # + G11/G12 sector labels
│   ├── top10_dft_validation.csv     # DFT feasibility of top-10 candidates
│   └── ring7_bnpah_predictions.csv  # 364 predicted 7-ring configurations
├── dft_inputs/                      # DFT input files for top-10
│   ├── orca/                        # ORCA .inp files
│   ├── gaussian/                    # Gaussian .gjf files
│   ├── structures/                  # XYZ coordinates (molecule + CO2 adsorption)
│   ├── run_orca_batch.sh            # ORCA batch submission script
│   └── run_gaussian_batch.sh        # Gaussian batch submission script
├── ring7_structures/                # Generated 7-ring isomers
│   ├── ring7_generated_isomers.csv  # 5 validated 7-ring SMILES
│   └── *.png                        # 2D structure images
├── figures/
│   ├── full_dossier_fig.png         # 16-panel comprehensive figure
│   ├── extended_validation_dft.png  # 6-panel DFT validation
│   ├── final_integrated_pipeline.png # 18-panel integrated figure
│   └── ring7_generated_isomers.png  # 7-ring isomer structures
├── docs/
│   ├── FINAL_DOSSIER_BNPAH_CO2_v0.38.md  # Main documentation
│   ├── FULL_DOSSIER_BNPAH_CO2_v0.36.md    # Extended validation
│   └── plant_analogy_formalized.md        # Plant photosynthesis analogy
├── scripts/
│   ├── bnp_compas_analysis.py       # Descriptor computation + GBR model
│   ├── co2_pipeline_final.py        # CO2 binding/transport/routing/yield
│   ├── dft_input_generator.py       # ORCA/Gaussian input generation
│   ├── generate_ring7_bnpah.py      # 7-ring BN-PAH generation
│   └── final_integrated_pipeline.py # Integrated analysis
├── tests/
│   └── test_descriptors.py          # Unit tests for descriptor computation
├── LICENSE                          # MIT License
├── CITATION.cff                     # Citation metadata (Zenodo)
└── README.md                        # This file

---

Key results

Top synthesis targets

#	Candidate	nDR	nSP	E_ads	capture_eff	Application
1	PAH61BN_0617	6	8	-67.0 kcal/mol	0.554	CO₂ capture
2	PAH26BN_0049	4	5	-41.0 kcal/mol	0.191	Methanol
3	PAH26BN_0005	1	0	-55.5 kcal/mol	0.468	Overall
4	PAH61BN_0616	6	6	-67.0 kcal/mol	0.554	Methane

Key descriptor statistics (23,856 molecules)

Descriptor	Pearson r with ΔHL	Ablation ΔMAE	Role
nDR	-0.84	+0.109 eV	PRIMARY — delocalization rating
nSP	-0.81	+0.042 eV	SECONDARY — separation pairs
nLL	-0.21	+0.013 eV	Useful
Bi/Bo, Ni/No	—	0	Redundant
nrings	—	-0.001	Redundant

nSP parity effect: t=34.3, p≈0 (even nSP → vinylic → higher HOMO-LUMO gap)

7-ring extrapolation (predicted)

nDR	E_ads (pred)	capture_eff (pred)	Improvement vs 6-ring
1	-36.3	0.407	+0.076
4	-43.4	0.456	+0.115
7	-56.5	0.619	+0.152

---

Methodological principles

1. No parameter fitting per molecule. All descriptors are universal graph properties.
2. Binding → Transport → Routing → Yield — full closed pipeline.
3. DFT-validated via semi-empirical MESP + Marcus barrier model (top-10 score=1.00).
4. Diatomic reference map (H₂ to C₂) validates bonding mechanism taxonomy.
5. Plant photosynthesis analogy — same topological features describe both.

---

Citation

If you use this pipeline, please cite:

BN-PAH CO2 Pipeline v0.38
Molecular Graph Descriptors for CO2 Capture, Transport, and Routing
https://github.com/YOUR_USERNAME/bnpah_co2_pipeline

See CITATION.cff for BibTeX.

---

Authors

Developed by Arena.ai Agent Mode.
Questions: open an issue on GitHub.

---

Contributing

1. Fork the repository
2. Create a feature branch (git checkout -b feature-name)
3. Run tests (pytest tests/)
4. Submit a pull request

License

MIT License — see LICENSE file.