Mathematical exploration and discovery at scale AlphaEvolve is a generic evolutionary coding agent that combines the generative capabilities of LLMs with automated evaluation in an iterative evolutionary framework that proposes, tests, and refines ...

A new paper with Bogdan Georgiev, Javier Gomez-Serrano, and Adam Zsolt Wagner: "Mathematical exploration and discovery at scale" arxiv.org/abs/2511.02864. Further discussion is at terrytao.wordpress.com/2025/11/05/m...

Gene Network Organization, Mutation, and Selection Collectively Drive Developmental Pattern Evolvability and Predictability The historical order in which mutations appear during evolution determine the evolutionary trajectory of gene regulatory networks and influence how developmental patterns change and diversify over tim...

Fresh from the press💥 We asked what happens when you evolve gene regulatory networks computationally at scale. Do general principles of GRN evolution jump out? Is the process predictable? Read on to find out @prxlife.bsky.social @crick.ac.uk @ucl-ipls.bsky.social 👉 journals.aps.org/prxlife/abst...

Proud to have contributed to this work! We built a stochastic agent-based model, revealing that axonal injury promotes retention in white matter leading to dense wild-type tumours, while absence of axonal injury leads to disseminated tumours and delayed glioblastoma progression.

We wrote this introductory review on how concepts from dynamical systems can help us understand developmental biology, hope it’s useful to somebody! It was a lot of fun to put together, and great to collaborate with brilliant colleagues across theory and experiments

Help me build a virtual version of my apartment by training a hugeee (like so huge) neural network on temperature data from my stove.

Call to action from the community to achive this ambitious goal!

A Potential Path to Safer AI Development Yoshua Bengio warns that the current approach to developing AI models carries potentially catastrophic risks.

Two years ago, I've completely reoriented my research to try to make AI safe by design. In this TIME op-ed, I present my team's new direction called "Scientist AI"; a practical, effective and more secure alternative to the current uncontrolled agency-driven trajectory.
time.com/7283507/safe...

Zero-shot evaluation reveals limitations of single-cell foundation models - Genome Biology Foundation models such as scGPT and Geneformer have not been rigorously evaluated in a setting where they are used without any further training (i.e., zero-shot). Understanding the performance of mode...

genomebiology.biomedcentral.com/articles/10....

Quite an indictment of some of the current single cell "virtual cell" foundation models. Even for the relatively mundane applications, cell labeling, batch correction etc, they are poor compared to much simpler & cheaper methods.

figure 3 (C) Representative images of LIT-1::GFP following the L2 asymmetric division in controls (WT), tissue-specific efl-3 mutants [efl-3(−)] and animals containing the lit-1(icb164) CRISPR deletion. Seam cell nuclei are circled in white. Hypodermal nuclei are labelled with blue arrowheads. a, anterior daughter cell; p, posterior daughter cell.

EFL-3/E2F7 modulates Wnt signalling by repressing the Nemo-like kinase LIT-1 during asymmetric epidermal cell division in C. elegans

Read this @reviewcommons.org transfer Research Article by @mar-ferrando-marco.bsky.social & Michalis Barkoulas @imperialcollegeldn.bsky.social
doi.org/10.1242/dev....

Protein degradation and growth dependent dilution substantially shape mammalian proteomes Cellular protein concentrations are maintained through a balance of synthesis and clearance. Clearance occurs through both protein degradation and growth-dependent dilution. At slow growth, clearance ...

Proteins degrade at different rates. These rates can span 2-3 orders of magnitude. When cells are not growing, this allows proteins that degrade slower than others to preferentially accumulate. This accumulation imparts influence on protein concentrations.
www.biorxiv.org/content/10.1...

Sure, happy to chat further.

Global transcription regulation revealed from dynamical correlations in time-resolved single-cell RNA sequencing This study presents a method for quantifying transcription dynamics from time-resolved single-cell RNA-sequencing data. The method integrates stochastic modeling with Bayesian inference to identify pa...

Very interesting work! Although we had a different inference approach, we similarly predicted peaks of degradation rate, predominantly in G1 & G1/S phase and peaks of transcription rate and burst frequency, predominantly in S phase and the late cell cycle: www.cell.com/cell-systems...