Chat with single cell
Bioinformatic agent
Intelligent Tool Selection
Bioinformatic agent
使用 MCP 构建 agent
构建你自己的生物信息学 agents
Agent 库
你可以在任何支持 MCP 的 Agent 库中使用 MCP 服务器。
如果你之前没有使用过任何 AI 库,你可以尝试 Agno(https://docs.agno.com/introduction)
这里是一个使用 scanpy-mcp 在 Agno 中构建 agent 的简单示例。
Copy
import asyncio
from agno.agent import Agent
from agno.tools.mcp import MCPTools
from agno.models.openai.like import OpenAILike
model = OpenAILike(
id="qwen-plus",
api_key="sk-**",
base_url="https://dashscope.aliyuncs.com/compatible-mode/v1",
extra_query={"enable_thinking": False},
)
async def run_agent(message: str) -> None:
"""使用给定的消息运行文件系统 agent。"""
async with MCPTools(
"scanpy-mcp run", timeout_seconds=60,
) as mcp_tools:
agent = Agent(
model= model,
tools=[mcp_tools],
show_tool_calls=True,
debug_mode=True,
description="""
你是一个生物信息学家。你擅长使用 Python 生物信息学工具,如 scanpy、anndata、pandas。
请逐个运行工具。
""",
monitoring=True
)
await agent.aprint_response(message, stream=True)
if __name__ == "__main__":
asyncio.run(
run_agent(
"read /data20T/dev/scmcphub/scanpy-mcp/tests/data/hg19; "
"then filter cells which gene number < 500; filter genes which express < 3 cells;"
"compute qc metrics,include mitochondrial and ribosomal genes, plot total counts, 线粒体百分比,gene number 的小提琴图, 质控图使用multiple_pannel,;"
"normalize and Logarithmize the data;"
"Identify highly-variable genes; Reduce the dimensionality, 绘制pca_variance_ratio;"
"perform cell clustering,设置resolution=0.5, 添加key leiden.0.5, "
"and draw UMAP and TSNE scatter plot, color by leiden.0.5, PTPRC, NKG7,KLRD1,GNLY,CST7,PRF1, 每行显示3个图;"
)
)
在此页面
助手
Responses are generated using AI and may contain mistakes.