AI models that learn information from genomic sequences have advanced prediction and design capabilities. Here we introduce Evo 2, a biological foundation model trained on 9 trillion DNA base pairs from a highly curated genomic atlas spanning all domains of life to have a 1 million token context window with single-nucleotide resolution.
Vitamins are essential metabolites that must be obtained from external sources. In modern times, they have become widely available, leading to their ad hoc consumption. We developed a nutritional genomics framework to systematically identify monogenic diseases responsive to micronutrient modulation.
The clinical success of cancer drug candidates depends on efficacy across many different individuals, but we currently rely on a limited set of in vivo preclinical models. Here, to address this limitation, we introduce GENEVA, a scalable single-cell-resolution platform for measuring responses to drug perturbations.
Cells integrate exogenous and endogenous signals, likely through dynamic functional associations between genes, but measuring these relationships is non-trivial. Here, we evaluate genetic associations in response to cell-cycle interruption, genotoxic perturbation, and nutrient deprivation using genetic interaction mapping in human cells.
We present MULTI-evolve, a rapid evolution framework that systematically engineers multimutants. Our approach combines protein language models or existing functional data with epistatic modelling to predict synergistic combinations.
High-altitude conditions improve glucose tolerance and reduce diabetes risk, but the physiological mechanism is not well understood. Using mouse models, we found that hypoxia alone robustly improved glucose tolerance and that the effect persisted for weeks after returning to normal oxygen levels.
Standard fixed-vocabulary tokenizers fragment biologically meaningful motifs, while nucleotide-level models preserve biological coherence but incur prohibitive computational costs for long contexts. We introduce dnaHNet, a state-of-the-art tokenizer-free autoregressive model that segments and models genomic sequences end-to-end.
Local hypoxia is a hallmark of solid tumors and a negative prognostic factor in the progression and treatment of cancer. Here, we showed that systemic hypoxia, in contrast to localized tumor hypoxia, decreases tumor growth in vivo across multiple cancer types and preclinical models.
Here, we report the development of RNA-guided trans-splicing with Cas editor (RESPLICE). RESPLICE uses two orthogonal RNA-targeting CRISPR effectors to co-localize a trans-splicing pre-mRNA and to inhibit the cis-splicing reaction, respectively.
From extrachromosomal DNA to neo-peptides, reprogramming of cancer genomes leads to the emergence of cancer state-specific molecules. Here, we systematically identify and characterize a large repertoire of orphan non-coding RNAs (oncRNAs), a class of cancer-emergent small RNAs, across 32 tumor types.