Background: Mass spectrometry (MS) is a very sensitive and specific method for protein identification, biomarker discovery, and biomarker validation. Protein identification is commonly carried out by comparing MS data with public databases. However, with the development of high throughput and accurate genomic sequencing technology, public databases are being overwhelmed with new entries from different species every day. The application of these databases can also be problematic due to factors such as size, specificity, and unharmonized annotation of the molecules of interest. Current databases representing liquid chromatography-tandem mass spectrometry (LC-MS/MS)-based searches focus on enzyme digestion patterns and sequence information and consequently, important functional information can be missed within the search output. Protein variants displaying similar sequence homology can interfere with database identification when only certain homologues are examined. In addition, recombinant DNA technology can result in products that may not be accurately annotated in public databases. Curated databases, which focus on the molecule of interest with clearer functional annotation and sequence information, are necessary for accurate protein identification and validation. Here, four cases of curated database application have been explored and summarized.
Findings: The four presented curated databases were constructed with clear goals regarding application and have proven very useful for targeted protein identification and biomarker application in different fields. They include a sheeppox virus database created for accurate identification of proteins with strong antigenicity, a custom database containing clearly annotated protein variants such as tau transcript variant 2 for accurate biomarker identification, a sheep-hamster chimeric prion protein (PrP) database constructed for assay development of prion diseases, and a custom Escherichia coli (E. coli) flagella (H antigen) database produced for MS-H, a new H-typing technique. Clearly annotating the proteins of interest was essential for highly accurate, specific, and sensitive sequence identification, and searching against public databases resulted in inaccurate identification of the sequence of interest, while combining the curated database with a public database reduced both the confidence and sequence coverage of the protein search.
Conclusion: Curated protein sequence databases incorporating clear annotations are very useful for accurate protein identification and fit-for-purpose application through MS-based biomarker validation.