Computational refinement of post-translational modifications predicted from tandem mass spectrometry

Area of Science:

• Proteomics
• Computational Biology
• Biochemistry

Background:

• Post-translational modifications (PTMs) are crucial for protein function and implicated in diseases.
• Analyzing tandem mass spectrometry data is key for genome-wide PTM discovery.
• Current blind PTM search methods suffer from noise and false positive assignments.

Purpose of the Study:

• To develop a machine learning algorithm, PTMClust, to enhance PTM prediction quality.
• To suppress noise and cluster peptides for improved PTM group identification.
• To reduce false PTM assignments and facilitate novel PTM discovery.

Main Methods:

• PTMClust algorithm applied to the output of blind PTM search engines.
• Clustering peptides with identical modifications.
• Evaluation using simulated datasets and real yeast proteome profiling data.

Main Results:

• PTMClust outperforms standard clustering algorithms and PTMFinder.
• Significant improvements in sensitivity and specificity across multiple PTM search engines.
• Reduced false PTM assignments, increased detection coverage, and discovery of novel PTMs, including terminus modifications.

Conclusions:

• PTMClust effectively refines PTM identification from mass spectrometry data.
• The algorithm aids in accurate PTM profiling and novel PTM discovery.
• PTMClust is a valuable tool for routine proteomic analysis.