Determination of enantiomeric composition of tryptophan by using fluorescence spectroscopy combined with backward interval partial least squares
Abstract
The application of backward interval partial least squares (BiPLS) method to fluorescence spectroscopy analysis was studied. A method which combines BiPLS and fluorescence spectroscopy was developed for determining the enantiomeric composition of tryptophan (Trp). Fluorescence spectroscopy was utilized to measure the interaction between Trp enantiomers and bovine serum albumin, which is the chiral selector of the two enantiomers. BiPLS was used to select spectral regions and build the calibration model. In terms of BiPLS, five spectral regions were selected and used to develop the calibration model between the spectral data and enantiomeric composition of Trp. In addition, a full-spectrum PLS model and two local-spectrum PLS models were developed in order to make a comparison to the BiPLS model. The prediction performance of the established models was assessed by external test validation and leave-one-out cross-validation. The BiPLS model shows the highest prediction accuracy among these models. For the BiPLS model, the root mean square relative error of external test validation and leave-one-out validation was 6.59% and 5.67%, respectively. It is demonstrated that fluorescence spectroscopy combined with BiPLS is a practicable method for determining the enantiomeric composition of Trp at trace levels. When there is 2.50 μmol L−1 Trp in the samples, the enantiomeric composition of Trp can be accurately determined. Furthermore, the result demonstrates that spectral region selection can significantly influence the fluorescence spectroscopy analysis and BiPLS is a practicable method for the spectral region selection in fluorescence spectroscopy analysis.