#35 · Tools & infrastructure

LLM agents and traceable toolkits are automating bioinformatics analysis workflows

The common thread is software architecture for automating and standardizing computational biology analyses. ToolsGenie 2.0 uses a ReAct-style multi-agent design with Docker-based tool selection to translate natural language queries into bioinformatics analyses. msmu, a Python package, addresses LC-MS proteomics post-search processing by organizing data in MuData objects with provenance logs that enable workflow replay. MCPmed, a community proposal, adapts the Model Context Protocol to expose bioinformatics web services, including GEO, STRING, and UCSC Cell Browser, to LLM agents. The dominant theme is LLM-driven orchestration layered over existing resources, with the proteomics toolkit contributing the traceability component. The items differ in maturity: two are preprints and one is a published proposal rather than an evaluated system, and no benchmark results appear in the item summaries.