Turbine

24/04/2026

We got very good at predicting biology. But can AI explain it?

Join us to explore the shift from black-box models to systems that reason, test hypotheses, and act as scientific partners 🧬

With speakers from Achira, NVIDIA, and Turbine. Click below to register.
https://eu1.hubs.ly/H0tPdm20

21/04/2026

We’re partnering with Crown Bioscience to bring together in silico scale with in vivo translatability.

Turbine’s Virtual Assays, combined with Crown’s tumor organoid platform, create a streamlined path from hypothesis to validation: helping R&D teams make faster, more confident decisions in translational oncology.

As our CEO & Co-founder Szabolcs Nagy puts it, this is about addressing a key trade-off in drug discovery between scale and translatability.

Read more: https://eu1.hubs.ly/H0tJtDg0

10/04/2026

The wet lab is an expensive place to find out that you were wrong. 💸

At , we’re showcasing how teams at AstraZeneca, Ono, and MSD are already working with us to predict more and test less. We use Virtual Assays first, simulating millions of perturbations, to find the fraction of experiments actually worth your time.

The data on our poster speaks for itself: 70% prospective hit rate, 50% shorter timeline to in vivo validation, 100% higher in vitro validation rate vs. industry benchmarks. 🚀

Abstract: https://eu1.hubs.ly/H0tp4xq0

Come see the results at our poster session on April 21 (9 AM - 12 PM) or visit booth #4845 to say hi to the team!

13/03/2026

We once tried to force our models to obey a thousand textbook biology rules. 📑

It made them worse.

Not because the model was wrong, but because the “general cell” those rules describe doesn’t actually exist.

A short cautionary tale about "Biological Truth Benchmarks" and overgeneralization.

Read more from our Founder & CTO Kristof Szalay:
https://eu1.hubs.ly/H0sFnHh0

or how your textbook lies

25/02/2026

Multimodality is a liability if your data isn't measuring the same underlying state. If your signals aren't aligned, you might just be injecting noise into the latent space instead of building a better representation.

Our rule of thumb: if modality A can’t help predict modality B, don't fuse them. Bigger isn't better, but alignment is.

Read the full breakdown from our Founder, CTO Kristof Szalay here:

What are we assuming in a multi-modal training? What can you measure?

24/02/2026

Big milestone!

We’ve closed a $25M Series B, led by Interactive Venture Partners with participation from Beiersdorf Venture Capital, a strong affirmation of confidence in our technology, our growing pharma traction, and the future of virtual biology.

We’re also moving beyond oncology, launching a new immunology collaboration with a (yet undisclosed) top 10 pharma company to model immune biology and rationally explore combination therapies at computational scale.

Biology is complex. Disease heterogeneity is massive.
We are virtualizating experiments to make it accessible.

Stay tuned: more pharma collaborations are in the making🔬

06/02/2026

Richard Izrael, PhD, Head of Lab, is at SLAS this week.

At Turbine, we use a lab-in-the-loop approach to virtualize biological experiments, creating virtual copies of wet-lab assays. We do this by:
👉 using agentic approaches to design assay protocols that form a reliable foundation for virtual assays
👉 running simulations to identify synergistic payload combinations before committing to physical experiments
👉 predicting outcomes with high confidence to make data generation faster and more efficient

The result: researchers can explore far more ideas at computational speed, while staying grounded in real experimental biology

If you’d like to dig into the details, reach out to us here on LinkedIn or catch Richard at the Networking Power Hour on Monday.

06/02/2026

This week invites reflection on progress in cancer research and on where new approaches may genuinely change how decisions are made.

ADCs continue to push precision medicine forward, particularly as the field explores more nuanced designs, including dual-payload strategies. These directions point to a broader shift moving beyond average responses toward a deeper understanding of biological context, heterogeneity, and trade-offs. That perspective aligns closely with this year’s World Cancer Day theme, United by Unique.

At Turbine, our work sits within this shift. By virtualizing biological experiments, we aim to support earlier exploration of complex questions from payload combinations to cell-line-specific effects before resources are committed in the lab. Recent efforts have focused on expanding biological coverage, improving how signals are surfaced across conditions, and exploring gene-level effects through virtual perturbations.

Equally important is the conversation around the science. Through our Virtual Biology Meetups, we learn alongside the community by sharing perspectives, pressure-testing ideas, and helping shape what comes next.

Progress in cancer research isn’t defined by a single breakthrough or a single day. It’s built incrementally, through better questions, better tools, and sustained collaboration.

Our latest note on dual-payload ADCs:
https://eu1.hubs.ly/H0rysMP0

Last Virtual Biology Community Meetup:
https://eu1.hubs.ly/H0ryrXP0

27/01/2026

Dual-payload ADCs won’t reach their full potential if they are designed by trial and error 🧪

Even as ADC pipelines are growing at record pace, tumor heterogeneity, resistance, and off-tumor toxicity remain major hurdles that often stand in the way of deeper and more durable clinical responses. Dual-payload ADCs are building a growing body of evidence showing how carefully combined cytotoxic mechanisms can help reach more tumor cells, slow the emergence of resistance, and move toward treatments that are both safer and more effective, ultimately offering more lasting relief for patients.

In this article, Jerzy Woźnicki, PhD discusses how Turbine’s Virtual Lab and in silico disease models can be used to screen and stress-test dual-payload combinations before they ever reach the bench, and how this approach helps uncover where the mechanisms behind true synergy and resistance are most likely to emerge. 🔬🖥️

Read the full article here 👇

Dual-payload ADCs are gaining traction, offering opportunities to overcome the tumour heterogeneity and resistance challenges faced by conventional ADCs.

20/01/2026

200+ live yesterday! 🚀

A huge thank you to everyone who joined our Virtual Biology Community Meetup. We couldn’t have asked for a better crowd to start our series.

The kick-off was all about the "first year of the virtual cell", discussing the challenge of virtualizing biological experiments and the road ahead for R&D.

Missed it or want to rewatch? Catch the full replay here: https://eu1.hubs.ly/H0r5gz20

This was just the first session. Keep your eyes peeled for more deep dives coming later this year, and follow our channels to stay updated:

🔹 X (Twitter): https://eu1.hubs.ly/H0r5fgG0
🔹 Bluesky: https://eu1.hubs.ly/H0r5c8B0
🔹 YouTube: https://eu1.hubs.ly/H0r5gBV0

​The past year marked a turning point for virtual biology and virtual cell modeling. After years of promises, bold claims, and accelerating hype, 2025 was th...

13/01/2026

🎉 The speaker lineup is complete!

Join us on January 19 for the Virtual Biology Community Meetup, a community-powered online event exploring where virtual cells and computational biology are heading in 2026.

We’re excited to welcome Fabian Theis, Director of the Institute of Computational Biology and Professor at TUM Mathematics and Life Sciences; Ron Alfa, MD, PhD, Co-founder & CEO at NOETIK; and Amaro Taylor-Weiner, PhD, former Chief AI Officer at Absci and Biotech Co-founder - moderated by Daniel Veres, MD, PhD, Co-founder & CSO at Turbine.

📅 Jan 19, 2026 | 4:30-6:00 PM CET
🌐 Online
👉 Register via Luma: https://eu1.hubs.ly/H0qXq-20

If you’re curious about the future of virtual biology, this is a great way to start the year after a busy JPM Week. 🚀

09/01/2026

High benchmark scores don’t mean your virtual cell is useful.

Impressive metrics often fail to translate into reliable biological insight to help decision-making. Benchmarks focused on limited real-world impact, weak or missing domain baselines, growing risks of data leakage, and overly convenient metrics can all create a false sense of progress.

In this post, our CTO and Gerold Csendes shares how we assess the actual utility of virtual cells in practice and why evaluation rigor, not leaderboard position, ultimately determines whether a model is worth trusting.

Read our note - Assessing a Virtual Cell’s Utility: https://eu1.hubs.ly/H0qQZ7T0
Substack version: https://eu1.hubs.ly/H0qQZNd0