Application of de Novo Sequencing to Large-Scale Complex Proteomics Data Sets

Area of Science:

• Proteomics
• Bioinformatics
• Mass Spectrometry

Background:

• Traditional protein identification relies on predefined sequence databases, limiting analysis of uncharacterized proteins.
• De novo peptide sequencing algorithms analyze mass spectra without databases but struggle with complex mixtures and result validation.
• Existing de novo sequencing methods lack robust validation and struggle with accuracy on longer peptides.

Purpose of the Study:

• To develop novel metrics for benchmarking de novo sequencing algorithms on large-scale proteomics data.
• To present a method for accurate false discovery rate calibration in de novo sequencing.
• To introduce a new algorithm (LADS) for enhanced de novo peptide sequencing accuracy and sensitivity.

Main Methods:

• Development of novel benchmarking metrics for de novo sequencing performance.
• Implementation of a false discovery rate calibration method for de novo results.
• Introduction of the LADS algorithm, utilizing experimentally disambiguated fragmentation spectra.

Main Results:

• Accurate calibration of false discovery rates for de novo sequencing results.
• The LADS algorithm demonstrates improved sequencing accuracy for longer peptides compared to existing methods.
• Enhanced discriminability between correct and incorrect peptide sequences identified via de novo sequencing.

Conclusions:

• Novel metrics and calibration methods improve de novo sequencing reliability.
• The LADS algorithm significantly enhances de novo peptide identification accuracy and sensitivity.
• This work enables accurate de novo identification of peptide sequences previously inaccessible through database search approaches.