Randomized sequence databases for tandem mass spectrometry peptide and protein identification.

Area of Science:

• Proteomics
• Bioinformatics
• Analytical Chemistry

Background:

• Tandem mass spectrometry (MS/MS) with database searching is standard for high-throughput peptide and protein identification.
• Current accuracy assessments like "high confidence" lack specificity and vary with experimental conditions.
• There is a need for robust, study-specific methods to estimate false positive rates in MS/MS identifications.

Purpose of the Study:

• To evaluate methods for estimating false positive identification rates in MS/MS data using randomized databases.
• To compare error rate estimations from randomized database searches with actual error rates from known protein standards.
• To provide a quantitative measure of peptide and protein identification accuracy.

Main Methods:

• Searches were performed using forward and randomized (reversed and reshuffled) databases.
• Separate and combined search strategies were examined.
• Methods were validated using MS/MS runs of known protein standards and applied to Shewanella oneidensis MR-1 samples.

Main Results:

• Randomized database searches provide quantitative estimates of false positive rates.
• Combined searches of a reshuffled database appended to a forward database were effective in estimating error rates.
• The proposed method allows for setting specific error rate thresholds.

Conclusions:

• Combined searches of reshuffled and forward databases are recommended for quantifying false positive rates in peptide and protein identification.
• This approach offers direct and quantifiable measures of accuracy, replacing vague assessments.
• Enables researchers to achieve desired error rates and enhances the reliability of proteomic data.