Red River Insights - Digital biology top10
where biology becomes software
Red River Insights - November 2025
Dear friends,
People keep asking what is next for AI. Some feel the slope has gone flat and they do not see impact. Yet the clearest answer is emerging in the life sciences: digital biology.
In a growing number of labs, biology is starting to behave more like an engineering discipline. Just as the semiconductor industry moved from laying out transistors by hand to using computer aided design tools, biology is having its own Computer-aided design (CAD) moment. We are moving from a hit-or-miss industry of discovery to an era of more predictable, programmable design.
The proof is already here. In the past few weeks, Eli Lilly partnered with NVIDIA to build one of the largest AI supercomputers ever owned by a pharma company, based on NVIDIA’s DGX systems, to invent new medicines faster. Google used an open model from its Gemma family to propose a new cancer-therapy pathway that was then validated at the bench. This is not just about human medicine. Using code and compute to design biological function, whether for a drug, a crop trait, or a sustainable material, instead of pure trial-and-error lab work, is quickly becoming the new operating standard for R&D.
What makes this shift exciting is how tangible it already feels. The path from an idea to a working biological function is compressing. Teams can design proteins and interactions in silico, explore millions of candidates, then send tighter, cleaner instructions into the lab. Models improved fast, with systems like AlphaFold 3 from DeepMind and ESM3 from former Meta researchers. Data became cheap and abundant, from omics (large-scale readouts of DNA, RNA and proteins) and single-cell sequencing. Compute and tools made the loop practical, from GPU stacks like NVIDIA’s BioNeMo (Biological Neural Models) to automated labs. This is AI meeting the real world in the most literal way: life.
All of this is why we shaped this edition around a theme we call Digital Biology. To fit in our theme, we distinguish between companies where biology is the product and companies where the digital or AI engine is the product. We focus on the latter.
Our latest RAMP ranking highlights the ten companies building these engines and the signals they send. Scroll down to discover who is turning the ultimate dream, making biology a software problem, into reality.
(unrelated to digital biology, but just as exciting: scroll down to read about the latest series A we led…)
RAMP's Digital Biology top10
We hope this sparks interesting conversations. If you have any comments or would like to suggest a startup that should be included, feel free to reach out to us. Joseph, Chloé and Olivier will be delighted to discuss these trends and rankings.
(Ranking established on 21/11/2025)
We’ve highlighted 4 key trends illustrated by these companies:
1. Generative Design: Asking The Computer For Molecules, Not Hits
Drug discovery used to mean screening millions of molecules and hoping one stuck. Generative design flips this around. Instead of asking “what works”, teams specify what they want and let AI search the space of chemistry and proteins.
Cradle is doing this for proteins. A scientist can tell its cloud platform “make this enzyme work at higher temperature” or “bind this target more tightly”. Cradle’s models then propose thousands of new protein variants, rank them, and help pick the few that are worth testing in the lab, like Copilot for protein engineers.
AQEMIA attacks the same problem from the physics side. Instead of training only on past experiments, they simulate how a drug and its target interact using quantum inspired algorithms, then steer generative models with those calculations. In plain terms, they can test billions of virtual molecules in the cloud and only bring the most promising ones into the wet lab.
Molecule.one solves the next bottleneck. Designing a clever molecule is useless if nobody can actually make it. Their platform uses AI to propose synthetic routes and reaction conditions, showing chemists how to go from a drawing on a slide to a real compound that can be manufactured. They combine software with a high throughput chemistry lab, so every digital suggestion is grounded in real chemistry.
Put together, these three companies sketch a future where the bottleneck in R&D is no longer ideas, but deciding which digitally designed molecules to advance.
2. Mapping The Source Code: Digital Twins Of Nature And Cells
If biology is a language, most of our current data covers only a few pages of the dictionary. The new TechBio engines are trying to read the rest.
Basecamp Research starts in the wild, not the lab. The team collects environmental samples from volcanoes, rainforests and other extreme ecosystems, then sequences the DNA to build what they call a digital twin of nature. This biodiversity dataset feeds AI models that can propose enzymes and biological parts that never appear in public databases. For companies designing drugs, materials or industrial strains, Basecamp is essentially expanding the search space by an order of magnitude.
DeepLife brings the same idea down to the level of individual human cells. The company trains deep learning models on multi omics data (integrated analysis of multiple biological data layers at once), learning how genes, proteins and pathways interact inside cells. The goal is to build digital twins of specific cell types, then simulate what happens if you perturb them with a drug, a gene edit or a combination of both. Instead of running thousands of physical experiments, researchers can explore scenarios in silico and carry only the most promising ones into the lab.
Relation Therapeutics embeds the patient directly into this loop. They combine single cell multi omics from real patient tissue with functional assays and machine learning, building detailed maps of diseases like fibrosis and osteoarthritis. Their platform is a “lab in the loop” system: models propose hypotheses, the lab tests them, and the results flow back to refine the models. It is an end to end discovery engine that treats drug design more like an iterative software project than a one shot scientific bet.
3. From Sequencers To Survivors: The New Clinical Bridge
Running clever models is not enough. To change care, they have to plug into real clinical workflows and real patients.
One Biosciences looks inside a tumor one cell at a time to see what each cell is doing.
They take standard biopsy samples (even old, preserved ones) and use advanced gene-reading tools to draw a very detailed “map” of each person’s cancer.
SeqOne Genomics operates one step earlier in the chain: making sense of sequencing data. Hospitals and labs use its SaaS platform to analyze next generation sequencing in oncology and rare diseases, automate workflows and generate clear clinical reports. Under the hood, SeqOne layers AI on top of complex genomic pipelines, so clinicians see an interpretable conclusion rather than raw variant files. As genomics spreads into routine care, this kind of “operating system” for sequencing is becoming mandatory infrastructure.
Cure51 flips the usual cancer question on its head. Instead of studying why most patients die from lethal cancers, they focus on the tiny fraction who survive far longer than expected in indications like metastatic pancreatic cancer or glioblastoma. Through the Rosalind Study, Cure51 is assembling tumor samples and clinical data from more than a thousand of these exceptional survivors across dozens of countries, then running them through a centralized multi omics and AI platform. The aim is to reverse engineer the biology of survival and turn those insights into new drug targets, signatures for patient selection and survival prediction tools that can plug into trials and care.
4. Biology On The Factory Floor
Digital biology is not only about medicines. Once biology becomes programmable, every industry that touches microbes or cells starts to look like an engineering problem.
Spore.Bio lives in that world. Today, quality control in food, beverage, cosmetics and pharma factories still relies heavily on petri dishes and slow cultures. Spore.Bio combines biophotonics and deep learning to detect microbes in real time, directly on the production line. Instead of waiting several days for a contamination result, customers can get a continuous microbial signal and act immediately. It is the difference between a smoke detector and a fire investigation report. As regulations tighten and brands become more sensitive to recall risks, this sort of always on bio sensing starts to look like core industrial infrastructure.
Open Questions:
Several important questions are still unresolved and matter for investors, founders and incumbents alike:
Should digital biology companies be software-first or fully integrated with wet labs?
Platforms like Cradle and Molecule.one lean toward being software-first, partnering with external labs rather than owning every experiment. Others like Relation Therapeutics, One Biosciences or DeepLife invest heavily in both computation and in-house or tightly coupled experimental systems. The open question is which model compounds faster over time and captures more of the value created.How should data rights and benefit sharing be structured in digital biology?
Basecamp Research is setting early norms by structuring commercial agreements with countries and partners where they collect biodiversity data. Cure51 does something similar with survivor cohorts and cancer centers worldwide. As these datasets become central to drug discovery and diagnostics, the unresolved question is how the industry should define clear standards on access, ownership and fair upside sharing.What will real moats look like in a world of open biology models?
Many biology foundation models are rapidly becoming more accessible. The defensibility for companies like Basecamp Research, DeepLife, One Biosciences or SeqOne Genomics likely sits in proprietary data pipelines, tightly integrated workflows and distribution into clinics or factories, not just in the models themselves. The open question is how strong those moats will prove to be as open models improve in the next cycle.
More on RAMP's scoring method
The ranking of these startups is based on the estimated momentum of the company, but the algorithm does not assess the quality or reliability of the products/solutions developed by these companies!
Find out about the algorithm behind this ranking and the way scores are calculated here: Cheat sheet on RAMP
In case you missed them, our latest top 10s are here: Women-founded startups (october25), Industry autonomy (September25) Advanced Materials (July25), Quantum (June25), Spanish tech (May25) Sport Tech (April 25), HR Tech (March25)
All the previous Top 10 are here.
Red River West leads Hyprspace series A!
Hyprspace is unlocking sovereign access to space with a patented hybrid propulsion engine (the fuel is solid and the oxidiser is liquid) that radically disrupts the economics of orbital mobility. By eliminating the need for turbopumps, their architecture is not only significantly cheaper and simpler but also inherently safer than traditional liquid propulsion systems.
Beyond commercial launch, this is a true dual-use breakthrough. HyPrSpace’s technology is uniquely positioned for defence applications, offering responsive launch capabilities critical for national sovereignty. Because their hybrid engines are storable, safe to handle, and quick to deploy, they open new doors for tactical applications, hypersonic systems, and reactive satellite deployment, ensuring rapid access to orbit when it matters most.
We are incredibly proud to back Alexandre, Nicolas, Sylvain and Vincent and their incredible team as they prepare for their first launch in 2026 and build a global leader in orbital mobility.
We are super happy to partner with DeepTech 2030 and The SPI fund from Bpifrance as well as Expansion Ventures
Other portfolio news:
The Exploration Company has acquired Thrustworks, a German additive manufacturing specialist, to expand its production capacity for high-performance space components and accelerate the industrialization of its Nyx spacecraft.
HyPrSpace has signed a Memorandum of Understanding with ATMOS Space Cargo to collaborate on future suborbital and orbital missions, combining French launch capabilities with sovereign European re-entry logistics using the Baguette-1 launcher.
Otera is partnering with Sopra Steria on a joint agentic deployment in the defense sector to prove that AI agents can operate safely, reliably, and at mission-critical scale in one of the most demanding environments on earth.
Worldia has opened a new office in Mérida, its third hub after Paris and Berlin, to anchor its growing travel-tech operations in Mexico and Latin America
Okeiro wrapped up an active presence at ASN Kidney Week 2025 in Houston, using five days of scientific sessions and clinical discussions to deepen collaborations with nephrologists and researchers and to showcase its precision-medicine platform for improving outcomes in kidney disease and transplantation.
You received this newsletter because we met or worked together and we thought you might be interested in an update, if we got it wrong, feel free to unsubscribe with the link at the end!