perfusio 0.1.0

Open reference implementation of self-driving perfusion bioprocess development (Gadiyar et al., 2026)

perfusio

perfusio is an open, peer-reviewable reference implementation of the self-driving perfusion methodology of Gadiyar et al. (2026) and Hutter et al. (2021), providing stepwise Gaussian Process hybrid models, entity-embedding transfer learning, 11 Bayesian experimental design acquisitions, and a fully-instrumented online digital twin for CHO cell perfusion bioprocesses.

---

Quick-start

pip install perfusio
python -c "from perfusio.simulator.cho_perfusion import CHOSimulator; print('OK')"

Reproduce all paper figures

pip install perfusio
python examples/reproduce_paper_figures.py
# → paper_figures/{fig4,fig7,fig8}.{pdf,png,svg}

CLI

# Generates 27 runs (4-factor Box-Behnken design, 3 centre points)
perfusio simulate --clone CloneX --n-days 28 --out runs/
perfusio train   --runs runs/ --model-out model.pt
perfusio run     --model model.pt --connector ambr250 --dashboard
perfusio reproduce-figures --out paper_figures/

---

What's inside

Module	Description
perfusio.mechanistic	CHO kinetics ODEs (dual-Monod, Pirt, Warburg, Luedeking–Piret)
perfusio.gp	PerfusionKernel, JackknifeEnsemble, StepwiseGP
perfusio.embedding	Entity-embedding transfer (Hutter 2021)
perfusio.hybrid	Hybrid state-space model + online retraining
perfusio.bed	11 BED acquisitions + Pareto / HV utilities
perfusio.simulator	CHOSimulator, Box-Behnken / LHC DoE, noise model
perfusio.twin	DigitalTwin, audit, notifications, scheduler
perfusio.connectors	OPC UA, SQL, filesystem, ambr®250 emulator
perfusio.metrics	rRMSE, PI coverage, CRPS, IGD+, ε-indicator
perfusio.viz	Static figures (Matplotlib) + interactive (Plotly/Dash)
perfusio.cli	Typer CLI

---

Installation

CPU (default)

pip install perfusio

GPU

pip install perfusio[gpu]

Dashboard

pip install perfusio[dash]
perfusio run --dashboard

Development

git clone https://github.com/lynchaos/perfusio.git
cd perfusio
pip install -e ".[dev]"
pre-commit install
pytest

---

Citation

If you use perfusio, please cite:

@article{gadiyar2026,
  author  = {Gadiyar, Chiraag J. and others},
  title   = {Self-Driving Development of Perfusion Processes
             for Monoclonal Antibody Production},
  journal = {Biotechnology and Bioengineering},
  year    = {2026},
  volume  = {123},
  number  = {2},
  pages   = {391--405},
  doi     = {10.1002/bit.28631},
}

@article{hutter2021,
  author  = {Hutter, Simone and others},
  title   = {Clone selection in cell culture development using
             multi-objective Bayesian optimisation with
             entity-embedding transfer learning},
  journal = {Computers \& Chemical Engineering},
  year    = {2021},
  volume  = {151},
  pages   = {107373},
}

---

Limitations

See LIMITATIONS.md for a full list of out-of-scope items, including CQA modelling, GMP validation, and proprietary media formulations.

Contributing

See CONTRIBUTING.md.

Code of Conduct

See CODE_OF_CONDUCT.md.

Maintainer

Kemal Yaylali — kemal.yaylali.uk

• GitHub: @lynchaos
• X / Twitter: @kmlyyll
• LinkedIn: kemalyaylali
• Support: [e-mail]

Licence

Apache-2.0 — see LICENSE.