In the era of large foundation models, insight is out-of-distribution !!
English
Suryanarayana Maddu
245 posts
Suryanarayana Maddu
@surimk92
Research fellow @FlatironInst Research Associate @Harvard MCB. PhD @mpicbg @csbdresden. Incoming assistant professor @JohnsHopkinsAMS @HopkinsDSAI
Katılım Nisan 2014
727 Takip Edilen387 Takipçiler
Suryanarayana Maddu retweetledi
1/12 Introducing MIMIC: a SOTA foundation model trained natively across DNA, RNA and proteins. MIMIC is multimodal and generative: it can use structure, regulation, evolution, and experimental context to infer missing biology or design new sequences.
English
@surimk92 nice! i just published this preprint; biorxiv.org/content/10.648… thoughts?
English
Based in San Carlos, Precigenetics was founded by hardware engineers, computational biologists, and cell biologists to make drug discovery truly relevant to human biology.
Today, I announce our 2026 plan: extending our measurement into a foundational dataset for cell states.🧵
English
@parmita Couldn’t agree more. This is exactly the direction we’ve been pushing from a biophysical perspective, trying to learn mechanistic models directly from time-resolved single cell datasets.
pnas.org/doi/abs/10.107…
arxiv.org/abs/2604.25062
English
Please share this with people who may be interested in such an effort.
We aim to make biology truly digitised across time, and when it comes to understanding disease, we believe we are truly at the start of a revolution.
It’s time to build.
English
English
🧵 1/10 Inferring mechanistic gene-regulatory programs from single-cell data in a principled, scalable way remains a central challenge in systems biology. Our new paper introduces Probability Flow Matching (PFM) to tackle it: 📄 arxiv.org/abs/2604.25062
English
10/10
Paper: arxiv.org/pdf/2604.25062
Code: vchz.github.io/pfi/
@FlatironInst
Grateful to @LucyIkkanda for help with the graphics.
English
9/10 PFM further extends to unbalanced cell populations, jointly inferring gene-regulatory and growth dynamics, and correctly assigns clonally tracked cells to correct fates.
English
Welcome to join us in building the next-generation agent framework, PantheonOS, to automate, scale, and accelerate scientific discovery, ultimately powering the Virtual Embryo platform.
Weize Xu@Nanguage
Huge congratulations to @Xiaojie_Qiu and the team on winning the Laude Institute Moonshots Seed Grant for the Virtual Embryo project. @PantheonOS will serve as the agent interface layer in this effort, helping connect AI agents with predictive model and large-scale multimodal embryogenesis data. Excited to be part of this vision: building a digital twin of mammalian embryogenesis and advancing the future of developmental biology and human health. 🚀🧬
English
I am looking for PhD students and postdoctoral researchers working at the broad themes of scientific machine learning, data-driven biophysical modeling, and the physics of learning. If these problems excite you, feel free to reach out !!
3/3
English
I would like to thank my family, friends, and mentors - @ivos99 ( @mpicbg ), @microbionaut, Mike Shelley ( @FlatironInst @SimonsFdn ), Dan Needleman( Harvard ) for their unwavering guidance and support.
2/3
English
I’m excited to share that I will be joining @JohnsHopkinsAMS as an Assistant Professor in Fall 2026. I will also be a member of the @HopkinsDSAI. 🎉🎉
1/3
English