FoodSpec

"Food decides the nature of your mind… Mind is born of the food you take."
— Sri Sathya Sai Baba, Effect of Food on the Mind, Summer Showers 1993 – Indian Culture and Spirituality (29 May 1993)

Production-ready Python toolkit for Raman/FTIR/NIR spectroscopy in food science. FoodSpec provides reproducible workflows for preprocessing, feature extraction, statistics, and machine learning, with built-in support for protocol-driven analysis, model management, and automated reporting.

Version: 1.0.0 | Status: Production Ready | Tests: 685 passing (79% coverage)

What problems does FoodSpec solve?

• Fragmented workflows: Vendor-specific formats, ad hoc preprocessing, irreproducible scripts.
• Lack of standards: No consistent data model across labs or instruments.
• Manual documentation: Time-consuming figure and report generation.
• Reproducibility challenges: Difficulty sharing, versioning, and archiving analyses.

What does FoodSpec provide?

Data & Import

• Unified data model for Raman, FTIR, and NIR spectroscopy
• CSV/TXT/JCAMP loaders; optional vendor support (SPC, OPUS)
• HDF5 spectral libraries for reference materials and calibration

Preprocessing

• Baseline correction (6 methods: ALS, rubberband, polynomial, airPLS, modified polynomial, rolling ball)
• Smoothing (Savitzky-Golay, moving average), normalization (vector, SNV, MSC)
• Derivatives, cropping, cosmic-ray removal
• ATR/atmospheric correction for FTIR

Feature Extraction & Interpretation

• Peak, band, and ratio detection with library-based chemical interpretation
• Ratiometric Questions (RQ) engine for reproducible ratio computation
• PCA/PLS scores, fingerprint similarity, VIP scores
• Feature importance summaries with visualization

Statistics & Quality Control

• Parametric and nonparametric hypothesis tests (t-test, ANOVA, MANOVA, Kruskal-Wallis)
• Bootstrap and permutation-based robustness
• Classification and regression metrics with confidence intervals
• Batch QC, novelty detection, drift monitoring, replicate consistency

Machine Learning

• Classification: Logistic, SVM, Random Forest, Gradient Boosting, PLS-DA
• Regression: Linear, Partial Least Squares, Random Forest
• XGBoost, LightGBM integration
• Optional Conv1D and MLP deep learning
• Model registry, versioning, and lifecycle tracking

Domain Workflows

• Oil authentication and quality assessment
• Heating/oxidation trajectory analysis
• Mixture composition estimation (NNLS, MCR-ALS)
• Hyperspectral imaging and spatial mapping
• Calibration transfer between instruments (Direct Standardization, PDS)

Reproducibility & Reporting

• Protocol-driven execution (YAML configuration)
• Automated narrative reports with metrics, tables, and figures
• Run metadata capture and artifact versioning
• Prediction confidence guards and quality gates
• Full provenance tracking

Supported modalities

Installation

Requires Python 3.10 or 3.11

pip install foodspec

# Optional extras
pip install 'foodspec[ml]'      # XGBoost, LightGBM
pip install 'foodspec[deep]'    # Conv1D, MLP deep learning
pip install 'foodspec[dev]'     # Documentation, tests, linting

See the [inAPI (5 minutes)

from foodspec import load_library
from foodspec.apps.oils import run_oil_authentication_workflow

# Load example dataset
fs = load_library("oils_demo.h5")

# Run complete authentication workflow
result = run_oil_authentication_workflow(fs, label_column="oil_type")

# Access results
print(f"Balanced Accuracy: {result.balanced_accuracy:.3f}")
print(f"Confusion Matrix:\n{result.confusion_matrix}")

CLI (5 minutes)

# Convert CSV to HDF5 library
foodspec csv-to-library raw_spectra.csv library.h5 \
  --wavenumber-col wavenumber --sample-id-col sample_id

# Run oil authentication workflow
foodspec oil-auth library.h5 \
  --label oil_type --output results/
📚 **[Full Documentation](https://chandrasekarnarayana.github.io/foodspec/)** | 150+ pages of guides, tutorials, and API reference

### Quick Links

- **Getting Started**
  - [Installation Guide](https://chandrasekarnarayana.github.io/foodspec/01-getting-started/installation/)
  - [Python Quickstart](https://chandrasekarnarayana.github.io/foodspec/01-getting-started/quickstart_python/)
  - [CLI Quickstart](https://chandrasekarnarayana.github.io/foodspec/01-getting-started/quickstart_cli/)
  - [15-Minute Tutorial](https://chandrasekarnarayana.github.io/foodspec/01-getting-started/quickstart_15min/)

- **User Guides**
  - [Data Libraries](https://chandrasekarnarayana.github.io/foodspec/04-user-guide/libraries/)
  - [CSV Import](https://chandrasekarnarayana.github.io/foodspec/04-user-guide/csv_to_library/)
  - [CLI Reference](https://chandrasekarnarayana.github.io/foodspec/04-user-guide/cli/)
  -Package Statistics

- **21,500+ lines** of production code
- **685 tests** with 79% coverage
- **150+ documentation pages** with examples
- **16 example scripts** + 3 Jupyter notebooks
- **10+ vendor formats** supported
- **6 baseline methods**, 4 normalization methods
- **10+ ML algorithms** with nested cross-validation
- **Protocol-driven workflows** for reproducibility
:

```bibtex
@software{foodspec2025,
  title = {FoodSpec: Pyt! Please:

- Read the [Contributing Guide](https://chandrasekarnarayana.github.io/foodspec/06-developer-guide/contributing/)
- Follow the [Documentation Guidelines](https://chandrasekarnarayana.github.io/foodspec/06-developer-guide/documentation_guidelines/)
- Write clear code with docstrings and type hints
- Add tests for new features (pytest)
- Ensure all checks pass: `pytest`, `ruff check`, `mkdocs build`

Open issues and pull requests with clear, concise descriptions. We appreciate bug reports, feature requests, and documentation improvement

See [CITATION.cff](CITATION.cff) for machine-readable citation metadata
## Development & Testing

```bash
# Run tests with coverage
pytest --cov=src/foodspec tests/ --cov-report=html

# Lint code
ruff check src/ tests/

# Build documentation
mkdocs build

# Run example scripts
python examples/phase1_quickstart.py

All tests passing ✅ | Coverage: 79% ✅ | Documentation builds ✅

• Advanced Topics

  • Model Registry
  • HSI & Harmonization
  • MOATS Implementation

• Developer Resources

  • Contributing
  • Documentation Guidelines
  • [Testing Coverage](https://chandrasekarnarayana.github.io/foodspec/06-developer-guide/testing_coverage

For more examples and tutorials, see the documentation.

Documentation

• Getting started: Installation
• Quickstart guides: Python • CLI • Protocols
• Data & IO: CSV import • Vendor formats • Libraries
• Preprocessing: Complete guide
• Features & Analysis: Feature extraction • Interpretation
• Workflows: Oil authentication • Heating analysis • Mixtures • Calibration
• ML & Statistics: Methods • Metrics
• Advanced: Protocols & automation • Registry & plugins • Deployment
• Reference: Glossary • API • Troubleshooting

Testing

pytest --cov          # Run tests with coverage report
ruff check            # Lint checks
mkdocs build          # Build documentation locally

Citation

If you use FoodSpec in your research, please cite the software. See CITATION.cff for full details.

Contributing

We welcome contributions. Before submitting, please:

• Follow guidelines in docs/contributing.md
• Write clear code with docstrings and examples
• Add tests for new features
• Ensure pytest, ruff, and mkdocs build pass

Open issues and pull requests with concise, clear descriptions.

Collaborators

• Dr. Jhinuk Gupta, Department of Food and Nutritional Sciences, Sri Sathya Sai Institute of Higher Learning (SSSIHL), Andhra Pradesh, India — LinkedIn
• Dr. Sai Muthukumar V, Department of Physics, SSSIHL, Andhra Pradesh, India — LinkedIn
• Ms. Amrita Shaw, Department of Food and Nutritional Sciences, SSSIHL, Andhra Pradesh, India — LinkedIn
• Deepak L. N. Kallepalli, Cognievolve AI Inc., Canada & HCL Technologies Ltd., Bangalore, India — LinkedIn

Author

• Chandrasekar SUBRAMANI NARAYANA, Aix-Marseille University, Marseille, France — LinkedIn

FoodSpec aligns spectroscopy, chemometrics, and ML into reproducible, well-documented pipelines for food science. Dive into the documentation for detailed theory, examples, and workflow guidance.