MSACL 2019 EU Abstract

Confident Protein and Peptide Identification with Mass Spectrometry: Understanding and Exploiting Collision Energy Dependence

Agnes Revesz (1), Tibor Andras Rokob (1), Daniel Szabo (1,2), Gitta Schlosser (2), Lilla Turiak (1), Karoly Vekey (1), Laszlo Drahos (1)
(1) Research Centre for Natural Sciences, HAS, Budapest, Hungary (2) Eotvos Lorand University, Budapest, Hungary

Warning: Undefined variable $headshot in /var/www/html/view_abstract/view_abstract_in_program.php on line 704

Agnes Revesz (Presenter) Research Centre for Natural Sciences, HAS >> POSTER (PDF)

Presenter Bio: I'm a chemist with expertise in various areas of mass spectrometry including modeling of gas phase processes, investigation of reaction mechanisms, ion mobility studies, complexation equilibria, and proteomics method development. Currently I'm a staff scientist at the Research Centre for Natural Sciences in Budapest, and my research is focused on understanding and exploiting properties of peptide fragmentation to support the development of more efficient proteomics workflows. Previously I held a postdoc position in the Institute of Organic Chemistry and Biochemistry in Prague. I obtained a PhD in Inorganic Chemistry from the Eotvos University in Budapest.

Relevant Financial Disclosures (within past 24 months) No relevant financial relationship(s) to disclose. Conflict mitigation NOT REQUIRED.

INTRODUCTION: Recently, MS-based proteomics has become a powerful tool in protein identification and characterization. Particularly for applications in a clinical setting where key information is to be obtained from samples of high complexity and variability, keeping up the pace with the ever-growing requirements is only possible via fully exploiting the potential in our instrumentation and data analysis workflows. In this respect, we recognized a lack of studies directly targeted toward achieving optimum proteomics identification and protein sequence coverage in bottom-up experiments.

OBJECTIVES: We want to explore how the choice of collision energy of the peptide MS/MS experiments affects high-level characteristics of results. We map the energy dependence of the confidence in database search identification (e.g., score), of the number of identified peptides/proteins, and of the sequence coverage for several types of instruments and search engines. We investigate the relationship between the optimal experimental settings and analyte characteristics to potentially serve as the basis of more advanced protocols.

METHODS: Tryptic digests of complex standards (HeLa, E. Coli) were subject to nano-HPLC-MS/MS experiments at dozens of different collision energies on Bruker QTOF and Thermo Orbitrap instruments. Spectra were matched against the SwissProt database using the Mascot and Byonic search engines. Further analysis, including comparison of LC-MS/MS runs and a nonlinear fitting of the score vs. collision energy curves, were performed using Scaffold 4 and our in-house developed software “Serac” (Search Engine Results Aggregation and Combination) processing search engine output files.

RESULTS: Our results revealed the complexity of parameter optimization by demonstrating that several peptides show bimodal score vs. energy behavior, due to the interaction of b/y fragment ion abundance patterns with the peptide identification confidence. The optimal collision energy, obtained as the peak positions of fitted Gaussians, follow linear trends with respect to m/z, but structural effects such as presence of mobile protons have significant effect. The importance of optimization is underlined by a 20-40% increase in the number of identified peptides and in the related sequence coverage, as well as a 20-30% increase in identification score, with respect to the factory default settings.

CONCLUSIONS: Our study is the first of its kind to address trends of and effects on collision energy dependence of a peptide identification score on a statistically significant set of peptides. The unprecedented bimodal behavior and the notable increases in score and identification efficiency indicate the usefulness of our studies in both deepening the understanding of proteomics workflows and in fine-tuning them for optimal performance.