genome-mcp

Instructions

Implementation Reference

• src/genome_mcp/core/tools.py:427-498 (handler)

  Core handler implementation for the 'smart_search' tool. Handles natural language queries, applies validation, filters, parsing, and execution via QueryExecutor. Adds smart_search_info to results.
  @mcp.tool() async def smart_search( description: str, context: str = "genomics", filters: dict[str, Any] = None, max_results: int = 20, ) -> SearchResult: """ 智能语义搜索 - 理解自然语言描述并执行相应查询 语义理解示例： - "breast cancer genes on chromosome 17" → 查找17号染色体上的乳腺癌基因 - "TP53 protein interactions" → 查找TP53蛋白相互作用 - "tumor suppressor genes" → 查找肿瘤抑制基因 - "genes related to DNA repair" → 查找DNA修复相关基因 Args: description: 自然语言描述 context: 搜索上下文（genomics/proteomics/pathway） filters: 过滤条件 max_results: 最大结果数 Returns: 智能搜索结果 Examples: smart_search("breast cancer genes on chromosome 17") smart_search("TP53 protein interactions", context="proteomics") smart_search("DNA repair genes", filters={"species": "human"}) """ try: # 验证搜索参数 validated_description, validated_context, validated_max_results = ( validate_search_params( description=description, context=context, max_results=max_results ) ) # 智能解析查询意图 query = _apply_filters(validated_description, filters) # 根据上下文调整查询 if validated_context == "proteomics": query_type = "protein" elif validated_context == "pathway": query_type = "search" else: query_type = "auto" # 解析查询意图 parsed = QueryParser.parse(query, query_type) # 执行查询（直接使用查询执行器，避免MCP工具间调用） result = await _query_executor.execute( parsed, max_results=validated_max_results ) # 添加智能解析信息 result["smart_search_info"] = { "description": validated_description, "context": validated_context, "parsed_query": query, "filters_applied": filters is not None, } return result except ValidationError as e: return format_simple_error(e, query=description, operation="smart_search") except Exception as e: return format_simple_error(e, query=description, operation="smart_search")
• src/genome_mcp/core/validation.py:184-217 (schema)

  Input schema validation function specifically for smart_search parameters: description (required str), context (genomics/proteomics/pathway), max_results (1-50).
  def validate_search_params( description: str, context: str = "genomics", max_results: int = 20 ) -> tuple[str, str, int]: """ 验证搜索参数 Args: description: 搜索描述 context: 搜索上下文 max_results: 最大结果数 Returns: 验证后的参数元组 Raises: ValidationError: 搜索描述验证失败 """ if not description or not description.strip(): raise ValidationError("Search description cannot be empty") description = description.strip() # 验证context valid_contexts = {"genomics", "proteomics", "pathway"} if context not in valid_contexts: context = "genomics" # 验证max_results if not isinstance(max_results, int) or max_results < 1: max_results = 20 elif max_results > 50: max_results = 50 return description, context, max_results
• src/genome_mcp/core/types.py:42-49 (schema)

  Output type definition (TypedDict) for smart_search results, including query, results list, total_count, and metadata.
  class SearchResult(TypedDict): """搜索结果类型""" query: str results: list[dict[str, Any]] total_count: int search_metadata: dict[str, Any]
• src/genome_mcp/main.py:34-41 (registration)

  Registration point where create_mcp_tools(mcp) is called, which defines and registers the smart_search tool using FastMCP's @mcp.tool() decorator.
  from .core.tools import create_mcp_resources, create_mcp_tools # 创建MCP实例 mcp = FastMCP("Genome MCP", version="0.2.5") # 注册所有资源和工具 create_mcp_resources(mcp) create_mcp_tools(mcp)