GPN-Star precomputed variant-effect scores for human genome and 5 model organisms land on Hugging Face
anshulkundaje · x · 2026-09-10
Song Lab at UC Berkeley released precomputed GPN-Star predictions on Hugging Face covering all possible SNVs in the human genome (hg38) plus five model organisms: mouse, chicken, fruit fly, C. elegans, and Arabidopsis.
- Massive scale: the main subset gpn-star-hg38-m447-200m-llr alone has 8.8B rows, with entropy and calibrated log-likelihood-ratio variants totaling in the tens of billions of rows
- Fields include chromosome, position, ref/alt alleles, and calibrated llr scores
- Distributed as Parquet under Apache-2.0, loadable via Dask/Polars for large-scale variant-effect screening
More from Research
- ProteinBlender: Free Open-Source Blender Addon for Publication-Quality Protein and DNA Animation — DeryaTR_ · 2026-09-10
- GPT-6 'Astra' Does 34 Math Steps in Latent Space, 4x More Than Sol — MaartenBaert · 2026-09-10
- Judea Pearl resurfaces his obscure 1978 paper on model complexity and credibility — yudapearl · 2026-09-10
- Q2D-Web corpus: nine domains, ten languages, English at 65.8% — perplexity_ai · 2026-09-10
- Fly connectome demos aren't what they seem: Beat Saber video is overfit replay, real RL not done yet — dotey · 2026-09-10
- Perplexity launches Q2D-Web, a benchmark and leaderboard for agentic RAG retrieval — perplexity_ai · 2026-09-10