Abstract
Background: Metabolite identification and annotation remain major bottlenecks in untargeted metabolomics because mass spectral features often lack sufficient specificity. High-confidence annotation requires experimental validation using authentic standards analyzed under matched chromatographic and ionization conditions, providing greater reliability than in silico predictions or database matching alone. This study aimed to develop a practical and scalable workflow for constructing a high-quality mass spectral library using a commercially available analytical standards kit. Methods: A total of 603 metabolites from the MSMLS kit were organized into 42 mixtures, each containing approximately 15 compounds. Mixture design was based on molecular mass and distribution coefficient values, specifically logD at pH 3.1, with a minimum logD spacing of 0.15 to improve chromatographic separation and reduce co-elution. This strategy was used to minimize the total number of injections while maintaining spectral quality. The resulting spectra were evaluated against online spectral resources and in silico fragmentation predictions. A preliminary proof-of-concept analysis was also performed using human serum samples. Results: Using this workflow, 471 metabolites, corresponding to approximately 78% of the standards, were successfully detected and incorporated into the spectral library. Comparison with online resources and in silico fragmentation predictions demonstrated improved spectral quality and reliability. The proof-of-concept serum analysis enabled identification of endogenous metabolites using the constructed library. In addition, the robustness and applicability of the workflow were further supported by a method validation study using metabolites derived from this library. Conclusions: This workflow provides a scalable strategy for constructing mass spectral libraries that balances spectral quality with analytical throughput. By using rational mixture design and authentic standards analyzed under matched experimental conditions, the approach enables substantial metabolite coverage while maintaining data reliability and minimizing experimental effort.