Google's AlphaGenomeAtlas maps predicted effects of all 9 billion possible DNA variants in a 1PB free database
AI寒武纪 · wechat · 2026-09-09
Google DeepMind launched AlphaGenomeAtlas, an AI-powered searchable database that precomputes the predicted molecular impact of every possible single-letter DNA change in the human genome — about 9 billion SNVs — into a 1PB resource, over 30x the size of the original AlphaFold database, free for academic research.
- AVI score: combines AlphaGenome and AlphaMissense to rank both coding and non-coding variants (98% of the genome is non-coding), with mechanistic attribution (e.g., disrupted RNA splicing, chromatin accessibility) and a library of 2,500+ frequent DNA motifs.
- Validated results: Broad Institute/GREGoR used it to find a previously missed DNM1 mutation linked to epileptic encephalopathy (experimentally confirmed); Exeter applied it to 54k+ UK Biobank genomes, finding 22% more non-coding genetic associations including regulators of PLA2G7 and EGLN1, plus 19 BMI-linked regions; Stowers used its motif set to classify transcription factors.
- Access: free web portal, a Google Antigravity skill integration, and commercial availability via Google Cloud; not for clinical use.
More from Research
- FrontierCS 2 launches on Kaggle: 48 tasks make agents beat published CS papers — MeganRisdal · 2026-10-27
- LightMTP: lightweight latent multi-token prediction boosts LLM planning in pretraining — gdm3000 · 2026-10-27
- DexPort: a handheld interface with a real robot hand for dexterous-hand data collection — pabbeel · 2026-10-27
- EMNLP study: AI literary translation reads fine but varies more, even within one book — tuvllms · 2026-10-27
- Lab monkeys at $30-80K each: a 4-year breeding cycle is the real preclinical drug bottleneck — zebird0 · 2026-10-27
- Custom VQA benchmark under 48GB VRAM: Qwen3.8-Flash-Next leads, Qwen3.8-27B regressed — EmilPi · 2026-10-27