Predicting cell state in previously unseen conditions such as disease or in response to a drug has typically required retraining for each new biological context. Today, Arc is releasing Stack, a foundation model that learns to simulate cell state under novel conditions directly at inference time, no fine-tuning required. Stack captures something that most models miss: cellular context. A T cell in inflamed tissue behaves differently, not just because of its own genes, but because of its environment. Stack processes cells together & learns from those relationships.

The genetic code is full of synonymous codons that, for decades, were assumed to be interchangeable.

Today, with @NVIDIA, co-led by @genophoria, we announce CodonFM, a family of open-source AI models that reveal the grammar underlying codon choice:
developer.nvidia.com/blog/introduci… There is a natural redundancy in the genetic code. 64 codons encode 20 amino acids, resulting in multiple "synonymous" codons encoding for the same amino acid.

Some of these, however, appear far more often than others, following consistent, non-random patterns across genes.

🧬Introducing a powerful, programmable new mechanism for genome design: DNA recombination with bridge RNAs.

As the first natural RNA-guided DNA recombinase, this system enables insertion, excision, or inversion of any two DNA sequences.

Out today in @Nature, led by Arc scientists @pdhsu, @mgdurrant, and @ntperry13 🧵

Read more:
Image by @visualsciencearcinstitute.org/news/blog/brid… 🪡CRISPR has revolutionized gene editing. But for true large-scale genome design, scientists need a precise and programmable way to rearrange large segments of DNA. The bridge recombinase - discovered in mobile genetic elements - joins any two DNA molecules in a modular way.