A universal database reduction method based on the sequence tag strategy to facilitate large-scale database search in proteomics

Abstract

Mass spectrometry-based metaproteomic and proteogenomic studies tend to use large-scale databases that may contain too many irrelevant or artificially constructed proteins. Such an imprecise database presents challenges for both the quality of peptide identification and the time consumption. To address them, we developed a database reduction method for iterative database searching, DBReducer, which can precisely and effectively reduce the large-scale database and is allowed to interface with any downstream database search engine. In addition, an entrapment strategy was introduced to evaluate the identification precision and recall of different search modes. Compared with the common one-step database search and the traditional iterative database search, the iterative search with DBReducer respectively improved the peptide identification recall from an average of 67.8% and 83.7%–93.5%, and respectively improved the peptide identification precision from an average of 91.1% and 89.6%–91.3%, and more importantly, using DBReducer respectively reduced the time consumption by an average of 57.7% and 68.2%. Our results indicate that DBReducer has the potential to be a widely used database reduction method prior to common proteomic analysis, especially for scenarios with large-scale databases.