Deep Learning Based Prediction of Gas Chromatographic Retention Indices for a Wide Variety of Polar and Mid-Polar Liquid Stationary Phases

Abstract

Prediction of gas chromatographic retention indices based on compound structure is an important task for analytical chemistry. The predicted retention indices can be used as a reference in a mass spectrometry library search despite the fact that their accuracy is worse in comparison with the experimental reference ones. In the last few years, deep learning was applied for this task. The use of deep learning drastically improved the accuracy of retention index prediction for non-polar stationary phases. In this work, we demonstrate for the first time the use of deep learning for retention index prediction on polar (e.g., polyethylene glycol, DB-WAX) and mid-polar (e.g., DB-624, DB-210, DB-1701, OV-17) stationary phases. The achieved accuracy lies in the range of 16–50 in terms of the mean absolute error for several stationary phases and test data sets. We also demonstrate that our approach can be directly applied to the prediction of the second dimension retention times (GC × GC) if a large enough data set is available. The achieved accuracy is considerably better compared with the previous results obtained using linear quantitative structure-retention relationships and ACD ChromGenius software. The source code and pre-trained models are available online.

A Deep Convolutional Neural Network for Prediction of Peptide Collision Cross Sections in Ion Mobility Spectrometry

“…It can be concluded that the 1D CNN gives an appropriate accuracy for the prediction of CCS values for peptides and can be used instead of the RNN or together with the RNN as part of an ensemble of models. An ensemble of independent models is usually more accurate than each of the models [ 19 , 20 , 23 ]. The averaging of the results of various models improves the accuracy [ 20 ].…”

“…An ensemble of independent models is usually more accurate than each of the models [ 19 , 20 , 23 ]. The averaging of the results of various models improves the accuracy [ 20 ]. Figure 3 demonstrates a correlation between the predicted and reference values and error distributions for different charge states.…”

“…Both architectures can also be used together. A 1D CNN was also applied for the prediction of the CCS of small molecules [ 18 ] and the prediction of gas chromatographic retention indices [ 19 , 20 ]. The position of features and single amino acids in a sequence severely affects CCS values [ 14 , 16 ].…”

Intelligent Workflow and Software for Non-Target Analysis of Complex Samples Using a Mixture of Toxic Transformation Products of Unsymmetrical Dimethylhydrazine as an Example

“…In criterion (vi), a larger threshold value is used compared with criterion (i): 100 units instead of 70. This is related to the lower accuracy of the RI prediction for the polar stationary phases, compared with the prediction for the non-polar stationary phases [30]. The previously published model [30] was used for prediction.…”

AIRI: Predicting Retention Indices and Their Uncertainties Using Artificial Intelligence

“…Stein et al used group additivity to estimate the RI values of diverse compounds passed through polar and nonpolar columns with mean absolute prediction errors of 46 and 65, respectively. The use of deep learning methods for improving upon QSAR has been a topic of investigation since the start of the deep learning revolution and has grown to encompass a large number of methods. , A number of these deep learning QSAR methods have been used to predict RI values ,− from the molecular structure with varying degrees of success.…”