Update Dockerfile

Dockerfile

@@ -43,35 +43,8 @@ RUN mkdir -p /data/transvar_db/refseq_hg38 /data/transvar_db/refseq_hg19
 # 如果设置了 HF_DATASET，从数据集下载
 RUN if [ -n "$HF_DATASET" ]; then \
     echo "Downloading from HF Dataset: $HF_DATASET" && \
-    python3 << 'PYEOF'
-import os
-from huggingface_hub import hf_hub_download, snapshot_download
-
-dataset = os.environ.get('HF_DATASET', '')
-target_dir = '/data/transvar_db'
-
-# 尝试使用 snapshot_download 下载整个目录
-try:
-    local_path = snapshot_download(repo_id=dataset, repo_type='dataset', cache_dir='/tmp/hf_cache')
-    # 复制到目标目录
-    os.system(f'cp -r {local_path}/* {target_dir}/')
-    print(f'Downloaded from Dataset: {local_path}')
-except Exception as e:
-    print(f'snapshot_download failed: {e}')
-    # 尝试单独下载文件
-    files = ['hg38/hg38.fa', 'hg38/ncbiRefSeq.txt.gz', 'hg19/hg19.fa', 'hg19/ncbiRefSeq.txt.gz']
-    for f in files:
-        try:
-            path = hf_hub_download(repo_id=dataset, filename=f, repo_type='dataset')
-            subdir = f.split('/')[0]  # hg38 or hg19
-            filename = f.split('/')[1]
-            os.makedirs(f'{target_dir}/refseq_{subdir}', exist_ok=True)
-            os.system(f'cp {path} {target_dir}/refseq_{subdir}/{filename}')
-            print(f'Downloaded: {f}')
-        except Exception as e2:
-            print(f'Failed to download {f}: {e2}')
-PYEOF
-fi
+    python3 -c "import os; from huggingface_hub import snapshot_download; dataset=os.environ.get('HF_DATASET',''); local=snapshot_download(repo_id=dataset, repo_type='dataset', cache_dir='/tmp/hf_cache'); os.system(f'cp -r {local}/* /data/transvar_db/')" || true; \
+    fi
 
 # 如果没有设置 HF_DATASET，从 UCSC 下载（首次构建）
 WORKDIR /data/transvar_db/refseq_hg38