Runchuan-BU/BioClaw

What it solves

BioClaw provides a conversational interface for bioinformatics research, eliminating the need for researchers to constantly switch between command-line tools, visualization software, databases, and literature search engines. It brings computational biology capabilities directly into common messaging apps like WhatsApp, WeChat, and Discord, allowing researchers to perform complex analyses from a chat interface.

How it works

Built on the NanoClaw container-based agent architecture and the STELLA framework, BioClaw uses a Node.js orchestrator to manage separate Docker containers for each conversation group. These containers come pre-installed with a comprehensive bioinformatics toolbox (including BLAST+, SAMtools, BWA, and PyMOL) and Python libraries (such as BioPython and pandas). The system is powered by the LLM (via Claude Agent SDK or OpenRouter) to translate natural language requests into tool executions. It also features a "Skills Hub" where the agent can dynamically fetch and execute specialized skills from a community repository.

Who it’s for

Bioinformatics researchers and wet-lab scientists who want to streamline their data analysis, sequence search, and literature mining without needing to extensive command-line expertise or switching environments.

Highlights

• Multi-Channel Integration: Supports WhatsApp, WeChat, Feishu/Lark, Discord, Slack, QQ, and a local web UI.
• Comprehensive Toolset: Integrated support for sequence analysis (BLAST, BWA), quality control (FastQC), structural biology (PyMOL), and data visualization (volcano plots, heatmaps).
• Dynamic Skill Acquisition: Can automatically fetch specialized skills from an external Skills Hub to handle tasks beyond its built-in capabilities.
• Chat-Native Controls: Includes built-in commands for workspace management, thread memory, and even host-side SSH execution directly from the chat.
• Wet-Lab Interpretation: Capable of analyzing uploaded gel/blot photos to assess lane quality and target bands.