CyChat: a conversational Cytoscape app for no-code, reproducible network analysis
Network-based analyses of molecular interactions are useful for interpreting high-throughput omics data and identifying therapeutic targets. Cytoscape is the standard platform for these tasks, but users face a trade-off between accessible graphical workflows that are difficult to document and reproducible automation in Python or R that requires programming expertise. General-purpose coding…
Cytoscape, the go-to platform for network-based analyses of molecular interactions, often presents a dilemma for users. On one hand, graphical workflows are user-friendly but lack reproducibility. On the other hand, Python or R automation scripts offer meticulous documentation and automation, yet demand programming skills. Efforts to bridge this gap with coding assistants have fallen short, existing as external tools.
Enter CyChat, a Cytoscape Desktop app that merges a conversational interface with a large language model (LLM) agent. This innovative tool converts natural language prompts into functional Cytoscape Automation workflows, executes any generated Python code, and exports chat sessions as standalone Jupyter notebooks. To streamline the setup process, CyChat incorporates an embedded Python runtime and accommodates both cloud-based and locally-hosted LLMs.
In rigorous testing, CyChat outperformed competitors, achieving a pass rate exceeding 99% across ten Cytoscape workflows using seven distinct LLM providers. In a real-world application, CyChat completed a network visualization task in 1.5-5 minutes, compared to an arduous 15-20 minutes for manual graphical workflows typically handled by computational biologists. CyChat is now available in the Cytoscape App Store, offering a no-code, reproducible solution for network analysis.
Written by urgent.news from bioRxiv's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
