Prediction of Ion Drift Times for a Proteome‐Wide Peptide Set Using Partial Least Squares Regression, Least‐Squares Support Vector Machine and Gaussian Process

Abstract

Quantitative structure-property relationships (QSPRs) have been developed to predict the ion mobility spectrometry (IMS) drift time t D for a set of 1481 peptides generated by protease digestion of the Drosophila melanogaster proteome using information directly derived from molecular structures. The relationship between peptide structure and the drift time t D was constructed by using partial least squares regression (PLS), least-squares support vector machine (LSSVM) and Gaussian process (GP) coupled with genetic algorithm-variable selection. Among these models, the linear PLS was incapable of capturing all dependences in this peptide system; nonlinear LSSVM and GP methods presented a good statistical performance on reproducing peptide mobility behavior. Moreover, since GP was able to handling both linear and nonlinear-hybrid relationship, it gave a stronger fitting ability and a better predictive power than the LSSVM. Systematic analysis of the GP model showed that diversified properties contribute remarkable effect to the relationship between the drift time and the peptide structure. Particularly, the structural topological information and charge distribution contribute significantly to the drift time of peptides.

Machine learning based prediction for peptide drift times in ion mobility spectrometry

“…The PLS approach is widely applied in the field of chemometrics (Wold et al, 2001), in sensory evaluation (Ortiz et al, 2006) and more recently to predict drift times in IMS (Liu et al, 2009). The PLS technique is a method for relating two matrices, X (the independent variable) and Y (the response variable) by a linear multivariate model, but goes beyond traditional regression in that it models also the structure of X and Y .…”

“…Prediction performance in a regression setting is calculated by selecting an error threshold and counting predictions within the error threshold as true predictions and those outside the threshold as false predictions. A more recent method proposed by Liu et al (2009) developed quantitative structure-property relationships (QSPR) to predict the drift times for a set of 1481 peptides using information derived from molecular structures. Additionally, they also investigated the use of multiple mathematical techniques such as partial least squares regression (PLSR), least-squares support vector machines (LS-SVM) (Suykens and Vandewalle, 1999) and a Gaussian Process (GP) to predict drift times.…”

Prediction of protein 13Cα NMR chemical shifts using a combination scheme of statistical modeling and quantum-mechanical analysis

“…Recently, QSPR approach has become very useful in the prediction and interpretation of several physical and chemical properties in the field of analytical chemistry. For example, it has been used in high performance liquid chromatography [25,26], ion mobility spectrometry [27][28][29], gas chromatography and gas chromatography-mass spectrometry [30][31][32], and so on. Related to NMR chemical shifts, Jurs et al used multiple linear regression and neural networks to predict 13 C NMR chemical shifts of several organic compounds [33][34][35][36].…”