iiinova — where your digital self keeps evolving

Valence Labs @valenceai.bsky.social · 17/04/2026

Want to learn more? Blog: valencelabs.substack.com/p/towards-re... Paper: arxiv.org/abs/2604.11661 Github: github.com/valence-labs...

valencelabs.substack.com

Towards reasoning in virtual cells

Bridging predict and discover with autonomous mechanistic explanations

000

Valence Labs @valenceai.bsky.social · 17/04/2026

Applying our VCR-Agent to the Tahoe-100M perturbation-cellular context pairs, we construct a dataset of 18,950 reasoning traces, called VCR-TRACES. We have publicly released this dataset on Github: github.com/yunhuijang/V...

100

Valence Labs @valenceai.bsky.social · 17/04/2026

We propose VCR-Agent: a multi-agent framework that generates structured reasoning traces from external biological sources such as StarkPrimeKG and Harmonizome. We use verifiers, like Boltz-2, for filtering and quality-control by evaluating the reliability of each step in the structured explanation.

100

Valence Labs @valenceai.bsky.social · 17/04/2026

Biological reasoning is often hard to verify since ground truth is scattered, context-dependent, and expensive to annotate. Inspired by pathway diagrams, we structure reasoning traces as DAGs of biological actions where each step can be independently falsified.

100

Valence Labs @valenceai.bsky.social · 17/04/2026

LLMs are strong at reasoning over unstructured text. But in biology, that reasoning is hard to verify. In our new preprint, we propose a framework for generating structured, falsifiable biological hypotheses from perturbation data. arxiv.org/abs/2604.11661

100

Valence Labs @valenceai.bsky.social · 17/04/2026

We propose VCR-Agent; a multi-agent framework that generates structured reasoning traces from external biological sources such as StarkPrimeKG and Harmonizome. We use verifiers, like Boltz-2, for filtering and quality-control by evaluating the reliability of each step in the structured explanation.

000

Valence Labs @valenceai.bsky.social · 16/04/2026

To realize our vision of better and more effective drug discovery, we must turn predictions into testable hypotheses and experimental designs. Dive into the Explain pillar of our Virtual Cell framework: why it matters and how we adapt LLM reasoning to biology. open.substack.com/pub/valencel...

000

Valence Labs @valenceai.bsky.social · 15/04/2026

Attending @iclr-conf.bsky.social in Rio? Interested in AI for drug discovery? We’d love to meet you. Valence Labs x LMRL Workshop are hosting a TechBio social event on April 26th. Rooftop bar, waterfront views, and engaging conversation. Register here: luma.com/9pyf9k90

000

Valence Labs @valenceai.bsky.social · 14/04/2026

Want to learn more? We've published the code: github.com/valence-labs... You can read the paper: arxiv.org/abs/2509.24779 Or come talk to us at ICLR: https://iclr.cc/virtual/2026/poster/10007880

github.com

GitHub - valence-labs/mars-fm

Contribute to valence-labs/mars-fm development by creating an account on GitHub.

000

Valence Labs @valenceai.bsky.social · 14/04/2026

Traditional MD requires long simulations in which interesting events are dominated by irrelevant dynamics. MarS-FM is our generative model that rethinks protein simulation by learning the transitions between biological states, achieving a 600x speedup. open.substack.com/pub/valencel...

100

Valence Labs @valenceai.bsky.social · 08/04/2026

Want to learn more? Come talk to us at ICLR! We will be presenting MarS-FM on April 24th at Poster Session 2. iclr.cc/virtual/2026...

iclr.cc

ICLR Poster MarS-FM: Generative Modeling of Molecular Dynamics via Markov State ModelsICLR 2026

000

Valence Labs @valenceai.bsky.social · 08/04/2026

Our MSM Emulators are a class of generative models that directly learn to sample the rare but key transitions between states. We instantiate this class with MarS-FM; using flow-matching, we outperform MIT's MDGen and Microsoft's BioEmu, while being 100x faster than MD.

100

Valence Labs @valenceai.bsky.social · 08/04/2026

MD is a powerful computational microscope, but its simulated trajectories contain high-frequency detail that is often irrelevant. In this work, we instead start from a coarse-grained representation that clusters conformations in distinct states using Markov State Models (MSM).

100

Valence Labs @valenceai.bsky.social · 08/04/2026

We're excited to announce that MarS-FM has been accepted at ICLR! We propose a new class of generative models that learn to sample state transitions of biomolecular systems, reproducing the statistics of Molecular Dynamics (MD) with drastic speedups. arxiv.org/abs/2509.24779

110

Valence Labs @valenceai.bsky.social · 29/10/2025

3/ We only have 8 presentation slots and sign-ups close on November 3rd. Make sure to RSVP below 👇 RSVP here: luma.com/n47bshdx

000

Valence Labs @valenceai.bsky.social · 29/10/2025

2/ We want to connect the people reimagining how drug discovery happens, to share ideas, spark collaborations, and explore ways to build together. If you want to present a “Show & Tell”, we review submissions on a rolling, first-come, first-served basis.

100

Valence Labs @valenceai.bsky.social · 29/10/2025

1/ Valence Labs is bringing together next generation of Montréal researchers and innovators in Bio-AI. We are hosting a Bio-AI Show & Tell — a quick and fun showcase of ideas, demos and projects in ML. RSVP here: luma.com/n47bshdx

101

Valence Labs @valenceai.bsky.social · 18/07/2025

5/ If you’re attending ICML, make sure to connect with @jhartford.bsky.social and Ihab to learn more about the research happening at Valence Labs and @recursionpharma.bsky.social.

000

Valence Labs @valenceai.bsky.social · 18/07/2025

4/ “Towards Scientific Discovery with Dictionary Learning: Extracting Biological Concepts from Microscopy Foundation Models” Link: arxiv.org/abs/2412.16247

100

Valence Labs @valenceai.bsky.social · 18/07/2025

3/ “A Cross Modal Knowledge Distillation & Data Augmentation Recipe for Improving Transcriptomics Representations through Morphological Features” Link: arxiv.org/abs/2505.21317

100

Valence Labs @valenceai.bsky.social · 18/07/2025

2/ ”ViTally Consistent: Scaling Biological Representation Learning for Cell Microscopy” Link: openreview.net/forum?id=niy...

100

Valence Labs @valenceai.bsky.social · 18/07/2025

1/ Valence Labs, @recursionpharma.bsky.social's AI research engine, is at ICML this week! Two of our scientists, @jhartford.bsky.social and Ihab Bendidi are presenting posters on work featured in their papers 👇🧵

110

Valence Labs @valenceai.bsky.social · 27/06/2025

3/ By combining SynFlowNet with Boltz-2, this new release aims to accelerate the design of high-affinity binders for more efficient early-stage drug discovery. Check out the GitHub repository here: github.com/recursionpha... Learn more about Boltz-2 here: www.rxrx.ai/boltz-2

000

Valence Labs @valenceai.bsky.social · 27/06/2025

2/ Originally introduced by Valence Labs, Recursion’s AI research engine, SynFlowNet helps ensure generative models propose molecules that are not only novel but also synthetically feasible.

100

Valence Labs @valenceai.bsky.social · 27/06/2025

1/ We are excited to open-source the SynFlowNet-Boltz-2 trainer today! 🧵

100

Valence Labs @valenceai.bsky.social · 25/06/2025

Pleased to have hosted the 2025 Molecular Machine Learning Conference last week at @mila-quebec.bsky.social Thank you to everyone who joined us and shared their valuable insights! We hope to see you all again soon.

020

Valence Labs @valenceai.bsky.social · 17/06/2025

Looking forward to seeing everyone at MoML this Wednesday, June 18th, at @mila-quebec.bsky.social in Montréal. We anticipate a day of engaging discussions, talks and poster presentations, exploring research at the intersection of machine learning and drug discovery. Agenda Here: portal.ml4dd.com

010

Valence Labs @valenceai.bsky.social · 06/06/2025

Check out some of the early media coverage here: www.forbes.com/sites/alexkn...

000

Valence Labs @valenceai.bsky.social · 06/06/2025

Proud to share @recursionpharma.bsky.social's exciting announcement on Boltz-2! Valence Labs, Recursion’s AI research engine, contributed to the foundational research supporting the @mit.edu team led by Regina Barzilay. Learn more about Boltz-2 here: www.rxrx.ai/boltz-2

120

Valence Labs @valenceai.bsky.social · 03/06/2025

3/ MoML is an excellent opportunity to engage with leading minds, explore cutting-edge research, and connect with the community in TechBio. We encourage everyone to secure their spot before it’s too late! Register now to be part of MoML 2025: portal.ml4dd.com/events/molec...

000

Valence Labs @valenceai.bsky.social · 03/06/2025

2/ Our list of speakers include: — Aaron Newman (@stanford.edu) — Smita Krishnaswamy (@yale.edu) — Gabriele Corso (@mit.edu) — Kirill Neklyudov (@mila-quebec.bsky.social) — Emmanuel Noutahi (@valenceai.bsky.social | @recursionpharma.bsky.social) — Michael LeVine (Genesis Therapeutics)

100

Valence Labs @valenceai.bsky.social · 03/06/2025

1/ We're excited to announce our speaker lineup for MoML 2025! Join us on June 18th, 2025, at Mila in Montreal for a day focused on the intersection of machine learning and drug discovery. 🧵

120

Reposted by Valence Labs

Recursion @recursionpharma.bsky.social · 22/05/2025

A new perspective paper from Recursion and our AI research engine @valenceai.bsky.social lays out our vision for a virtual cell as a system that can reliably drive the discovery of new drugs via an iterative loop of: predict, explain, discover. arxiv.org/abs/2505.14613 #TechBio 🧪

041

Valence Labs @valenceai.bsky.social · 22/05/2025

5/ TxPert addresses the "Predict" capability of our virtual cell framework This represents an initial, but important, step in a much larger journey: building models that can predict, explain, and eventually discover new therapeutic opportunities www.valencelabs.com/advancing-dr...

valencelabs.com

Advancing Drug Discovery Outcomes with Virtual Cells at Recursion - Valence Labs

Predict, Explain, Discover: The Pillars of the Virtual Cell Our view of the virtual cell rests on three interconnected capabilities:

110

Valence Labs @valenceai.bsky.social · 22/05/2025

4/ We’re also releasing the TxPert App, a tool that makes the model accessible to researchers - Select a cell type and gene to perturb - Visualize the gene interaction networks - Analyze predicted expression changes Explore more: txpert.valencelabs.com

130

Valence Labs @valenceai.bsky.social · 22/05/2025

3/ TxPert achieves state-of-the-art performance by integrating multiple biological knowledge networks These include curated public resources like STRINGdb and GO, alongside @recursionpharma.bsky.social proprietary PxMap and TxMap graphs derived from large-scale perturbation screens

120

Valence Labs @valenceai.bsky.social · 22/05/2025

2/ Observing how cells respond to perturbations is key to understanding disease and designing more effective therapies But doing this experimentally across an intractable combination of genes, cell types, and conditions is slow, expensive, and infeasible

110

Valence Labs @valenceai.bsky.social · 22/05/2025

1/ Introducing TxPert: a new model that predicts transcriptional responses across diverse biological contexts It’s designed to generalize across unseen single-gene perturbations, novel combinations of gene perturbations, and even new cell types 🧵 www.valencelabs.com/txpert-predi...

valencelabs.com

TxPert: Predicting Cellular Responses to Unseen Genetic Perturbations - Valence Labs

We introduce TxPert: a state-of-the-art model that leverages multiple biological knowledge networks to accurately predict transcriptional responses under OOD scenarios.

121

Valence Labs @valenceai.bsky.social · 20/05/2025

7/ This work represents an important step towards our mission of decoding biology to radically improve lives We are excited to share our vision and invite the scientific community to explore these resources together

110

Valence Labs @valenceai.bsky.social · 20/05/2025

6/ By leveraging multiple biological knowledge graphs, TxPert can generalize to unseen single perturbations and combinations of novel cell types and multiple perturbations Read the paper: www.valencelabs.com/publications...

110

Valence Labs @valenceai.bsky.social · 20/05/2025

5/ As an important first step toward the Predict capability of this framework, we are also introducing TxPert: a state-of-the-art model for predicting transcriptional responses to genetic perturbations, especially in out-of-distribution scenarios

120

Valence Labs @valenceai.bsky.social · 20/05/2025

3/ These principles guide our progress toward virtual cells: mechanistic models of functional response that can enable therapeutic discovery Read the blog: www.valencelabs.com/advancing-dr...

valencelabs.com

Advancing Drug Discovery Outcomes with Virtual Cells at Recursion - Valence Labs

Predict, Explain, Discover: The Pillars of the Virtual Cell Our view of the virtual cell rests on three interconnected capabilities:

110

Valence Labs @valenceai.bsky.social · 20/05/2025

2/ This framework structures our approach to developing models that can: 🔬 Predict cellular responses across diverse biological contexts, timepoints, & modalities 🔍 Explain these responses by identifying key interactions & pathways 💡 Discover new therapeutic opportunities

110

Valence Labs @valenceai.bsky.social · 20/05/2025

1/ At Valence Labs, @recursionpharma.bsky.social's AI research engine, we’re focused on advancing drug discovery outcomes through cutting-edge computational methods Today, we're excited to share our vision for building virtual cells, guided by the predict-explain-discover framework 🧵

2136

Valence Labs @valenceai.bsky.social · 13/05/2025

(3/3) Join us this year on June 18th at @mila-quebec.bsky.social. Tickets are free for students. Register today: portal.ml4dd.com/events/molec...

000

Valence Labs @valenceai.bsky.social · 13/05/2025

(2/3) Last year, MoML brought together 180+ attendees, 59 universities, and 21 companies for a full day of talks, posters, and discussions on the latest advances of machine learning and drug discovery. Watch the 2024 recap video: www.youtube.com/watch?v=7Xyo...

youtube.com

MoML at Mila 2024 Recap

YouTube video by Valence Labs

100

Valence Labs @valenceai.bsky.social · 13/05/2025

(1/3)The poster submission deadline for MoML 2025 has been extended to May 20th, 2025. Don’t miss an opportunity to share your work at this years conference. Submit here: portal.ml4dd.com/moml-2025-po...

111