# levacells

> Levacells uses artificial intelligence and genome-scale metabolic modelling to design custom cell culture media and feed supplements for biopharma and biotech. We help teams improve cell growth, titre, and product quality with far fewer wet-lab experiments than traditional screening.

Levacells SRL is a spin-off of NOVA University Lisbon, incorporated in Brussels, Belgium. Our technology is grounded in peer-reviewed hybrid genome-scale modelling and culture media engineering from the NOVA / LAQV REQUIMTE group. The final formulation recipe belongs to the client — they can use it, modify it, and choose any manufacturer.

## Pages

- [Home](https://levacells.com/): Company overview — custom AI-designed cell culture media and feed supplements.
- [Technology](https://levacells.com/technology): Five-step pipeline from genomic data to optimized media, feeds, and feeding strategy. Includes the five peer-reviewed publications underpinning the platform.
- [About](https://levacells.com/about): Company origins (NOVA University Lisbon spin-off), team bios, and partner logos.
- [Contact](https://levacells.com/contact): Book a discovery call via Proton Calendar, send a message via Tally form, or email hello@levacells.com directly.
- [Privacy Policy](https://levacells.com/privacy): GDPR-compliant privacy policy for Levacells SRL.

## What Levacells does

Levacells designs custom cell culture media and feed supplements using a proprietary hybrid AI pipeline. The service covers:

1. **Genome & annotation** — curate genomic data for the target cell line (CHO, HEK, and other mammalian lines with publicly sequenced genomes).
2. **Genome-scale model (GEM)** — build or adapt a metabolic model capturing metabolites, reactions, and transport specific to the medium and cell line.
3. **Designed experiments** — targeted design of experiment (DoE) to measure exchange fluxes and omics readouts with a minimal number of wet-lab runs.
4. **Hybrid semi-parametric AI** — blend mechanistic flux balance analysis (FBA) with data-driven constraints (PCA / ML) so predictions respect both biology and process data.
5. **Optimize & deliver** — multi-objective optimization proposes media composition, feed supplement, and feeding strategy; validated in-lab before delivery.

## Key capabilities

- Custom media and feed supplement composition tailored to a specific cell line and process goals
- Genome-scale metabolic modelling grounded in published science
- Multi-objective optimization (cell growth, titre, glycosylation profiles, physicochemical properties)
- Tailor-made feeding strategies compatible with existing bioreactor setups
- Orders-of-magnitude fewer experiments than traditional combinatorial screening
- Client owns the final recipe — no licensing, no lock-in

## Differentiation

Levacells is different from traditional media optimization in three key ways:
- **Science-grounded**: predictions are constrained by a genome-scale metabolic model, not purely statistical correlations.
- **Multi-objective**: optimize multiple targets simultaneously (e.g. improve titre while controlling glycosylation).
- **Experiment-efficient**: hybrid AI reduces wet-lab burden by orders of magnitude compared to classical Design of Experiment or high-throughput screening.

## Target customers

- Biopharma and biotech companies producing therapeutic proteins, monoclonal antibodies, or viral vectors using mammalian cell culture.
- CDMOs and CROs that want to offer media optimization as a service.
- Academic groups transitioning cell culture processes from research to manufacturing scale.

## Team

- **Eliot Boulanger** — CEO. LinkedIn: https://www.linkedin.com/in/eliot-boulanger-81468713/
- **Rui Oliveira** — Senior Scientific Advisor. Co-founder of the NOVA LAQV REQUIMTE systems biotechnology group, originator of the genome-scale metabolic modelling and functional enviromics methodology. LinkedIn: https://www.linkedin.com/in/rui-oliveira-17470614/
- **João Ramos** — Lead Scientist. Developed the hybrid semi-parametric FBA framework for CHO cell culture media design. LinkedIn: https://www.linkedin.com/in/jo%C3%A3o-ramos-18144769/

## Research foundation (selected publications)

- Ramos et al. (2022) "Genome-scale modeling of Chinese hamster ovary cells by hybrid semi-parametric flux balance analysis" — Bioprocess and Biosystems Engineering. https://doi.org/10.1007/s00449-022-02795-9
- Pinto et al. (2023) "Hybrid deep modeling of a CHO-K1 fed-batch process: Combining first principles with LSTM neural networks" — Frontiers in Bioengineering and Biotechnology. https://doi.org/10.3389/fbioe.2023.1237963
- Ramos et al. (2020) "A dynamic model linking cell growth to intracellular metabolism and extracellular by-product accumulation" — Biotechnology and Bioengineering. https://doi.org/10.1002/bit.27288
- Oliveira et al. (2014) "Culture media engineering by projection to latent pathways: The case of Pichia pastoris" — Biotechnology and Bioengineering. https://doi.org/10.1002/bit.25332
- Oliveira et al. (2010) "A cell functional enviromics method for the engineering of cell culture media" (patent / MediaOmics spin-off). https://patentsencyclopedia.com/app/20140017706

## Company information

- **Legal name**: Levacells SRL
- **Founded**: Brussels, Belgium (NOVA University Lisbon spin-off)
- **Address**: 8 Avenue de la Renaissance, 1000 Brussels, Belgium
- **Email**: hello@levacells.com
- **LinkedIn**: https://www.linkedin.com/company/levacells/
- **Website**: https://levacells.com