Data from: Near-infrared spectroscopy for metabolite quantification and species identification

Near-infrared spectroscopy (NIRS) is a high-throughput method to analyse
the near-infrared region of the electromagnetic spectrum. It detects the
absorption of light by molecular bonds and can be used with live insects.
In this study, we investigate the accuracy of NIRS in determining
triglyceride level and species of wild caught Drosophila. We employ the
chemometric approach to produce a multivariate calibration model. The
multivariate calibration model is the mathematical relationship between
the changes in NIR spectra and the property of interest as determined by
the reference analytical method. Once the calibration model was developed,
we used an independent set to validate the accuracy of the calibration
model. The optimized calibration model for triglyceride quantification
yielded coefficients of determination of 0.73 for the calibration test set
and 0.70 for the independent test set. Simultaneously, we used NIRS to
discriminate two species of Drosophila. Flies from independent sets were
correctly classified into D. melanogaster and D. simulans with accuracy
higher than 80%. These results suggest that NIRS has the potential to be
used as a high throughput screening method to assess a live individual
insect’s triglyceride level and taxonomic status.