FamDB database download – Jie Hua’s Bioinformatics Notes

# 设置 URL 列表
URLS=(
"https://www.dfam.org/releases/current/families/FamDB/dfam39_full.0.h5.gz"
"https://www.dfam.org/releases/current/families/FamDB/dfam39_full.0.h5.gz.md5"
"https://www.dfam.org/releases/current/families/FamDB/dfam39_full.1.h5.gz"
"https://www.dfam.org/releases/current/families/FamDB/dfam39_full.1.h5.gz.md5"
"https://www.dfam.org/releases/current/families/FamDB/dfam39_full.7.h5.gz"
"https://www.dfam.org/releases/current/families/FamDB/dfam39_full.7.h5.gz.md5"
)

# 使用 aria2c 多线程下载
for url in "${URLS[@]}"; do
    echo "Downloading $url ..."
    aria2c -x 16 -s 16 -k 1M "$url"
done
# 校验 MD5
for file in *.gz; do
    echo "Checking MD5 for $file ..."
    md5file="${file}.md5"
    if [ -f "$md5file" ]; then
        # 格式转换为标准 md5sum 可读格式
        hash=$(cut -d ' ' -f 1 "$md5file")
        echo "$hash  $file" > check.md5
        md5sum -c check.md5
        if [ $? -ne 0 ]; then
            echo "❌ MD5 check failed for $file"
            exit 1
        else
            echo "✅ MD5 check passed for $file"
        fi
        rm check.md5
    else
        echo "⚠️ Warning: MD5 file not found for $file"
    fi
done
# 解压缩
for file in *.gz; do
    echo "Unzipping $file ..."
    gunzip -k "$file"
done
echo "✅ All files downloaded, verified, and unzipped successfully."