hong-niu/Variant-Foundation-Embeddings

本数据集提供了基于高质量功能性数据（来自FAVOR、ClinVar和GWAS Catalog的注释）的大规模遗传分析所用的遗传变异水平嵌入向量。这些嵌入向量使用OpenAI的text-embedding-3-large模型生成，适用于HapMap3和MEGA数据集中的1.5百万个遗传变异。每个遗传变异通过其染色体、位置（基于hg38版本）、参考等位基因和替代等位基因来标识，并配有LLM嵌入向量。

This dataset provides variant level embeddings for large-scale genetic analysis based on high-quality functional data from annotations by FAVOR, ClinVar, and GWAS Catalog. These embeddings are generated using OpenAIs text-embedding-3-large model for 1.5 million genetic variants in the HapMap3 and MEGA datasets. Each genetic variant is identified by its chromosome, position (based on hg38), reference allele, and alternate allele, along with the LLM embedding vector.