遇见数据集

AetherXeno: in-silico saturation mutagenesis atlas of liver xenobiotic-receptor (PXR/FXR/AhR) non-coding regulatory variant effects

收藏
Zenodo2026-06-19 更新2026-06-21 收录
官方服务:

资源简介:

# AetherXeno — in-silico saturation mutagenesis atlas of liver xenobiotic-receptor regulatory variant effectsTri-receptor (PXR / FXR / AhR) atlas of non-coding regulatory variant effects for inductionpharmacogenomics. AlphaGenome-finetuned per-receptor effect scores (`ft_head_delta`) for**17,267,831 variants × 3 receptors**:- **14,853,408** = complete in-silico saturation of the receptor peak union (every possible SNV)- **2,414,423** = gnomAD-observed SNVs in ATAC-extended regionsScored on an H200 (4 GPU, bfloat16); cross-platform QC vs an RTX-5090 replicate r=0.999 (PXR/FXR),0.998 (AhR). Model-independent receptor-grammar validation: high-score variants concentrate on knownRXRA/NR1I2 motifs (Fisher OR 15.1); per-base saturation sensitivity 1.57–1.91× higher on motif cores.## Files- `aetherxeno_combined.parquet` — full 17.27M atlas (variant_id, chrom, pos, ref, alt, per-receptor ft_head_delta + normalized + effect_direction, gnomAD AF, region_class).- `variant_xref.parquet` — the 14.85M receptor-confirmed layer with cross-annotation (GTEx liver eQTL, ClinVar, GWAS Catalog, FIMO motif, phyloP).- `tierA_eqtl_validated.parquet`, `tierB_rare_regulatory.parquet` — high-priority tiers.- `saturation_motif_profile.json`, `QC_REPORT_h200.json` — validation + QC.- `DATA_DICTIONARY.md` — full field/coverage documentation (incl. honest rsID/observed-vs-novel scope).Interactive query: the full 17.27M atlas is queryable at the AetherXeno web app, Supabase-backed.License: CC-BY-NC-4.0, matching AlphaGenome non-commercial terms.

提供机构:
Zenodo
创建时间:
2026-06-19
二维码
社区交流群
二维码
科研交流群
商业服务