PeakSelect: preprocessing tandem mass spectra for better peptide identification

Abstract

We present a new preprocessing method, PeakSelect, to improve the accuracy and efficiency of Tandem Mass-Spec peptide (protein) identification. The fundamental difference between noise and fragment ions in spectra is that ions have isotopes but noise does not. We propose a new and important concept of an Isotope Pattern Vector (IPV) which characterizes the isotope cluster of fragment ions. Then the noise and real peaks can be distinguished by the quantitative IPV values. PeakSelect first uses a new method of the Gaussian Mixture Model and Expectation-Maximization (EM) algorithm to find the base intensity level (baseline) in a spectrum. Then PeakSelect selects features based on the IPV and baseline, and constructs a decision tree to automatically classify the peaks into different categories such as noise, single ion peaks, and overlapping peaks. Experiments show that PeakSelect can help to reduce the Mascot searching time and increase the reliability of peptide identifications. In particular, PeakSelect performs well on complex spectra with a large number of peaks from large peptides, and supports more sequence identification than other well-known systems.

Isotope pattern vector based tandem mass spectral data calibration for improved peptide and protein identification

“…Overlapping peaks are difficult to separate. We resolve this issue by following a previously reported method 32. Essentially, we summarize the overlapping peaks into several predominant types, and modify the above definitions of normalized deviations.…”

“…Dataset C contains 114 tandem spectra selected from the production of a Q‐TOF Ultima Global spectrometer for tryptic digestion peptides of eight proteins: Myoglobin (horse skeletal muscle), BSA (bovine serum albumin), fetuin (fetal calf serum type III), lysozyme (egg white), alpha‐lactoalbumin (bovine), BCA (bovine milk), phosvitin (egg yolk), and ribonuclease B (bovine pancreas) 32. All three datasets were used to illustrate the performance of IPV‐based systematic measurement error prediction.…”

Deconvolution in mass spectrometry based proteomics

“…Least absolute shrinkage and selection operator (LASSO)‐based variable selection models have been developed for simultaneously deconvoluting peptides with overlapping isotopic patterns and arbitrary set of charge states. In the PeakSelect system, a decision tree is constructed to classify the noise peaks, single and overlapped ion peaks in quadrupole time‐of‐flight mass spectra using features such as isotopic pattern vector, assigned charge of peak and isotopic L‐score peak number in the cluster …”

Features-Based Deisotoping Method for Tandem Mass Spectra

“…And, also, isotopic peaks can overlap that could result in wrong interpretation of masses of fragment ions. Thus, to increase the accuracy of the peptide identification and reduce the complexity of MS/MS analysis, many existing deisotoping algorithms [4–19] have already been explored to detect the isotopic clusters of fragment ions.…”

“…Some of these deisotoping methods [4–10, 19] are based on the theoretical isotopic distribution matching with experimental isotopic distribution. And the theoretical isotopic distribution can be estimated according to the monoisotopic mass of peptide ions [5, 17, 20, 21].…”