Evo 2 (Arc Institute, NVIDIA)
A DNA language model trained on 9.3 trillion bases that can flag cancer-causing BRCA1 variants without being told about them.
Evo 2 (2025; 7B and 40B parameters, 1M-token context) predicts variant effects zero-shot, including BRCA1 pathogenicity, and generates genomic sequences; open weights.
How it works
StripedHyena long-context sequence model over nucleotides.
- Zero-shot variant effect
- Open
- Human regulatory prediction weaker than specialised models
Latest papers
topQuery for this technology: (TITLE:"Evo 2" OR ABSTRACT:"Evo 2" OR TITLE:"Arc Institute, NVIDIA" OR ABSTRACT:"Arc Institute, NVIDIA") AND (cancer OR tumor OR tumour OR oncology OR carcinoma OR lymphoma OR leukemia OR leukaemia OR myeloma OR sarcoma OR melanoma OR glioma). Results are unfiltered search hits about Evo 2 (Arc Institute, NVIDIA), not a curated reading list.
Pages like this
not linked directly; found by shared links- TechnologyAlphaGenome
Shares the tags foundation-model, genome.
- TechnologyAlphaMissense
Shares the tags foundation-model, genome.
- TechnologyEnformer and Borzoi (DeepMind, Calico)
Shares the tags foundation-model, genome.
- TechnologyNucleotide Transformer (InstaDeep)
Shares the tags foundation-model, genome.
- TechnologyGenePT
Shares Stanford Health Care / Stanford Cancer Institute and the tag foundation-model.
- TechnologyState (Arc Institute perturbation model)
Shares Arc Institute and the tag foundation-model.
- TechnologyMerlin (Stanford abdominal CT vision-language model)
Shares Stanford Health Care / Stanford Cancer Institute and the tag foundation-model.
- TechnologyMUSK (Stanford, vision-language pathology)
Shares Stanford Health Care / Stanford Cancer Institute and the tag foundation-model.