U.S. Department of Energy

Pacific Northwest National Laboratory

Elimination of systematic mass measurement errors in liquid chromatography-mass spectrometry based proteomics using regression models and a priori partial knowledge of the sample content.

TitleElimination of systematic mass measurement errors in liquid chromatography-mass spectrometry based proteomics using regression models and a priori partial knowledge of the sample content.
Publication TypeJournal Article
Year of Publication2008
AuthorsPetyuk VA, Jaitly N, Moore RJ, Ding J, Metz TO, Tang K, Monroe ME, Tolmachev AV, Adkins JN, Belov ME, Dabney AR, Qian W-J, Camp DG, Smith RD
JournalAnal Chem
KeywordsAlgorithms, Calibration, Chromatography, Liquid, Complex Mixtures, False Positive Reactions, Mass Spectrometry, Models, Biological, Peptides, Protein Processing, Post-Translational, Proteomics, Regression Analysis, Reproducibility of Results, Sensitivity and Specificity, Trypsin
Abstract

The high mass measurement accuracy and precision available with recently developed mass spectrometers is increasingly used in proteomics analyses to confidently identify tryptic peptides from complex mixtures of proteins, as well as post-translational modifications and peptides from nonannotated proteins. To take full advantage of high mass measurement accuracy instruments, it is necessary to limit systematic mass measurement errors. It is well known that errors in m/z measurements can be affected by experimental parameters that include, for example, outdated calibration coefficients, ion intensity, and temperature changes during the measurement. Traditionally, these variations have been corrected through the use of internal calibrants (well-characterized standards introduced with the sample being analyzed). In this paper, we describe an alternative approach where the calibration is provided through the use of a priori knowledge of the sample being analyzed. Such an approach has previously been demonstrated based on the dependence of systematic error on m/z alone. To incorporate additional explanatory variables, we employed multidimensional, nonparametric regression models, which were evaluated using several commercially available instruments. The applied approach is shown to remove any noticeable biases from the overall mass measurement errors and decreases the overall standard deviation of the mass measurement error distribution by 1.2-2-fold, depending on instrument type. Subsequent reduction of the random errors based on multiple measurements over consecutive spectra further improves accuracy and results in an overall decrease of the standard deviation by 1.8-3.7-fold. This new procedure will decrease the false discovery rates for peptide identifications using high-accuracy mass measurements.

DOI10.1021/ac701863d
Alternate JournalAnal. Chem.
PubMed ID18163597
PubMed Central IDPMC2518823
Grant ListP01 HL066973-05 / HL / NHLBI NIH HHS / United States
P41 RR018522-05 / RR / NCRR NIH HHS / United States
RR18522 / RR / NCRR NIH HHS / United States
| Pacific Northwest National Laboratory