LLM-Assisted Development of a Locally Deployable Molecular Networking Toolkit: Enabling Customizable Analysis in Natural Products

MN-Suite is an open-source, locally deployable molecular networking toolkit developed through LLM-assisted software engineering to provide a flexible, server-independent workflow for natural product MS/MS analysis. The toolkit integrates six similarity algorithms and three spectral modes (MS2, neutral loss (NL), and hybrid MS2+NL), offering a customizable GUI-based framework for local preprocessing, network construction, and visualization. In the Aconitum data set examined here, the neutral-loss entropy-similarity strategy (NL-ES) produced the highest internal RCF score among the tested algorithm-data combinations (RCF = 0.537). By combining diagnostic-ion/neutral-loss filtering with a seed-neighborhood strategy, the MN-Suite prioritized 26 putative alkaloid analogues for further structural confirmation. These results support MN-Suite as a practical local workflow for configurable molecular networking and illustrate how domain experts can use LLM-assisted software engineering under human oversight to develop specialized computational tools.