boltz_mcp:基于 Boltz2 的蛋白质建模 MCP 服务器项目

可用于预测蛋白质3D结构、蛋白质-配体结合亲和力及管理相关作业。该项目提供13种工具,支持异步作业监控、批量处理、预计算MSA加速筛选及输入验证,通过Docker快速部署。【此简介由AI生成】

分支1Tags0
文件最后提交记录最后更新时间
6 个月前
8 个月前
5 个月前
5 个月前
6 个月前
5 个月前
5 个月前
5 个月前
5 个月前
6 个月前
6 个月前

Boltz MCP 服务器

通过 Docker 使用 Boltz2 进行结构和亲和力预测

一款适用于 Boltz2 蛋白质建模的 MCP(模型上下文协议)服务器,具备 13 项工具功能:

  • 从序列或 YAML 配置预测蛋白质 3D 结构
  • 预测蛋白质-配体结合亲和力
  • 提交长时间运行的作业并进行异步监控
  • 跟踪和检索作业结果
  • 验证蛋白质序列和配体 SMILES
  • 批量处理多个序列
  • 支持预计算 MSA,实现快速配体筛选

Docker 快速入门

方法 1:从 GitHub 拉取预构建镜像

# Pull the latest image
docker pull ghcr.io/macromnex/boltz_mcp:latest

# Register with Claude Code
claude mcp add boltz -- docker run -i --rm \
  --user `id -u`:`id -g` --gpus all --ipc=host \
  -v ~/.boltz:/opt/boltz_cache:ro \
  -v `pwd`:`pwd` \
  ghcr.io/macromnex/boltz_mcp:latest

方法 2:本地构建 Docker 镜像

git clone https://github.com/MacromNex/boltz_mcp.git
cd boltz_mcp
docker build -t boltz_mcp:latest .

claude mcp add boltz -- docker run -i --rm \
  --user `id -u`:`id -g` --gpus all --ipc=host \
  -v ~/.boltz:/opt/boltz_cache:ro \
  -v `pwd`:`pwd` \
  boltz_mcp:latest

要求:

  • 支持 GPU 的 Docker(nvidia-docker 或带有 NVIDIA 运行时的 Docker)
  • Boltz2 模型检查点位于 ~/.boltz(首次本地运行时自动下载)
  • 已安装 Claude Code

Docker 标志说明

标志 用途
-i MCP 标准输入输出的交互模式
--rm 退出后移除容器
--user `id -u`:`id -g` 以当前用户身份运行(输出文件归您所有)
--gpus all 访问所有 GPU
--ipc=host 用于 PyTorch 性能的主机 IPC 命名空间
-v ~/.boltz:/opt/boltz_cache:ro 挂载缓存的模型检查点(只读)
-v path:path 挂载工作目录以进行输入/输出访问

用于快速筛选的预计算 MSA

当针对同一蛋白质筛选多个配体时,挂载预计算的 MSA 目录可跳过缓慢的 MSA 服务器查询(每次运行节省约 5 分钟):

claude mcp add boltz -- docker run -i --rm \
  --user `id -u`:`id -g` --gpus all --ipc=host \
  -v ~/.boltz:/opt/boltz_cache:ro \
  -v /path/to/msa_files:/opt/msa:ro \
  -v `pwd`:`pwd` \
  boltz_mcp:latest

然后将 msa_path="/opt/msa/protein.a3m" 传递给任何预测工具。MSA .a3m 文件会在首次使用 --use_msa_server 运行时自动生成,可在输出目录的 msa/ 文件夹下找到。


验证安装

claude mcp list
# You should see 'boltz' in the output

可用工具(共 13 个):

  • simple_structure_prediction / simple_affinity_prediction — 同步预测
  • submit_structure_prediction / submit_affinity_prediction / submit_batch_structure_prediction — 异步任务
  • get_job_status / get_job_result / get_job_log / cancel_job / list_jobs — 任务管理
  • validate_protein_sequence / validate_ligand_smiles — 输入验证
  • list_example_data — 列出捆绑示例

使用示例

快速结构预测

Predict the structure of this protein: MVLSEGEWQLVLHVWAKVEADVAGHGQDILIRLFKSHPETLEKFDRFKHLKTEREQESRKEAAG. Save to /path/to/results/.

蛋白质-配体亲和力

Predict binding affinity between protein MVLSE... and aspirin (SMILES: CC(=O)Oc1ccccc1C(=O)O). Use simple_affinity_prediction.

使用预计算MSA进行快速配体筛选

Screen these 10 ligands against KRAS using simple_affinity_prediction with msa_path="/opt/msa/kras.a3m" for each.

具有潜力的高质量结构

Submit a structure prediction with use_potentials=True for the sequence in /path/to/protein.fasta. Monitor until complete.

命令行界面使用方法(无 MCP)

# Structure prediction
./env/bin/python scripts/structure_prediction.py --sequence "MVLSE..." --output results/out

# Affinity prediction
./env/bin/python scripts/affinity_prediction.py \
  --protein-seq "MVLSE..." --ligand-smiles "CC(=O)O" --output results/out

# With pre-computed MSA (skips MSA server, much faster)
./env/bin/python scripts/affinity_prediction.py \
  --protein-seq "MVLSE..." --ligand-smiles "CC(=O)O" \
  --msa-path /path/to/protein.a3m --output results/out

故障排除

未找到 Docker?

docker --version  # Install Docker if missing

GPU 无法访问?

  • 确保已安装 NVIDIA Docker 运行时
  • 使用 docker run --gpus all ubuntu nvidia-smi 命令进行检查

模型检查点缺失?

  • 先运行一次本地预测,模型会自动下载到 ~/.boltz
  • 或在运行前设置 BOLTZ_CACHE=/path/to/cache

Claude Code 未找到?

npm install -g @anthropic-ai/claude-code

后续步骤

有关可用工具、YAML 配置格式、本地设置和性能优化的全面指南,请参见 detail.md

许可证

MIT

项目介绍

可用于预测蛋白质3D结构、蛋白质-配体结合亲和力及管理相关作业。该项目提供13种工具,支持异步作业监控、批量处理、预计算MSA加速筛选及输入验证,通过Docker快速部署。【此简介由AI生成】

定制我的领域