Key Investor Signals
Signed LOIs and active biopharma contract talks
Lemna Bio reports signed LOIs and active contract discussions with leading biopharma companies for fee-for-service projects and early partnerships.
Repeat founder with Y Combinator track record
CEO Moustafa Houmani co-founded Adaptyv Biosystems, a protein-engineering startup that joined Y Combinator and raised a $2.5M pre-seed round led by Wingman Ventures in 2022.
CHF 400,000 secured in non-dilutive support
Lemna Bio secured a CHF 100,000 EPFL Innogrant, CHF 150,000 from Venture Kick, a CHF 100,000 FIT Tech Seed loan and a CHF 50,000 DayOne NEXT grant.
Other Positive Signals
- PeSTo (2023) and CARBonAra (2024) were both published in Nature Communications, with Lucien Krapp as first author.
- Four soluble CARBonAra-designed TEM-1 variants were highly thermostable and retained catalytic activity.
- Selected from 194 applications for the 2025 DayOne Accelerator cohort.
- Lemna Bio reports 3,000+ PeSTo users and says Novo Nordisk ranked CARBonAra among two leading protein-design models.
Team
Moustafa Houmani
Co-founder & CEO
- Earned an MSc in Bioengineering from EPFL; his ETH Zürich thesis focused on light-inducible transcription factors.
- Co-founded Adaptyv Biosystems, a protein-engineering startup built around high-throughput protein synthesis and screening.
- Managed EPFL’s 2018 iGEM team, which won a Gold Medal and received two category nominations.
Lucien Fabrice Krapp
Co-founder & CTO
- First author of CARBonAra (Nature Communications, 2024), a context-aware protein-design model validated on β-lactamase enzymes.
- Completed his PhD and postdoctoral research at EPFL’s Laboratory for Biomolecular Modeling, where he developed Lemna Bio’s core technology.
- First author of PeSTo (Nature Communications, 2023), a geometric transformer for predicting protein binding interfaces.
Prof. Matteo Dal Peraro
Scientific co-founder & CSO in residence
- Senior author of the PeSTo and CARBonAra papers that underpin Lemna Bio’s core modelling approach.
- Leads EPFL’s Laboratory for Biomolecular Modeling, combining computational and experimental structural biology with molecular engineering.
- Directs EPFL’s Institute of Bioengineering and has been an EPFL faculty member since joining as a tenure-track professor in 2007.
Product
Lemna Bio builds physics-grounded AI models that work directly from molecular structure to predict interactions and support early drug-discovery decisions. The platform maps binding sites, assesses target druggability, and designs protein sequences or molecular probes for experimental testing.
Geometric transformer core
The models operate directly on atomic coordinates and element types, without hand-crafted molecular descriptors. This lets a single framework analyse proteins and nearby molecules across diverse structural contexts.
Two peer-reviewed models
PeSTo maps protein interfaces with proteins, nucleic acids, ions, ligands and lipids. CARBonAra proposes sequences for a protein backbone in its molecular context; experiments produced four stable, catalytically active enzyme variants.
Multiple routes to biopharma
Lemna Bio plans to combine fee-for-service projects with early biopharma partnerships. It also develops custom molecular probes so partners can test model-generated hypotheses in the lab, while preparing a platform for direct model access.
Market Overview
Drug development is costly and attrition is severe: Deloitte estimates that the average cost of progressing a drug from discovery to launch reached $2.67 billion in 2025. This makes earlier target and mechanism decisions economically important, particularly before companies commit heavily to chemistry, preclinical work and clinical trials.
Pharmaceutical and biotechnology companies are the primary buyers for this type of technology. Comparable platforms monetise through research collaborations, model licensing, subscription access and fee-for-service projects. Grand View Research estimates that these buyers generated 59.2% of AI drug-discovery market revenue in 2025.
Roughly 1 in 10 drug programmes entering clinical development ultimately reaches approval.
Insufficient efficacy accounts for an estimated 40–50% of clinical drug-development failures.
Pharma and biotech companies generated 59.2% of AI drug-discovery market revenue in 2025.
The global AI drug-discovery market is forecast to grow from $2.3B in 2025 to $13.8B by 2033.
Comparable Outcomes
Nimbus Therapeutics
Nimbus used structure-based computational chemistry and predictive modelling to develop NDI-034858, an oral TYK2 inhibitor for immune-mediated diseases.\n\nTakeda acquired the subsidiary holding NDI-034858 in February 2023 for $4B upfront, with two $1B payments tied to annual sales milestones.
Isomorphic Labs
Isomorphic Labs uses AI models to support small-molecule drug discovery and design through multi-target pharmaceutical collaborations.\n\nIn January 2024, Lilly and Novartis paid a combined $82.5M upfront under collaborations offering up to $2.9B in potential milestones, plus royalties.
Interested in More Startups Like Lemna Bio?
Get a curated pipeline of healthcare and life sciences startups aligned with your investment thesis.
Sources
This profile was built from public company materials, ecosystem sources, market references, and professional profiles.