High-Throughput Quantitative Biochemical Characterization of Algal Biomass by NIR Spectroscopy; Multiple Linear Regression and Multivariate Linear Regression Analysis

One
of the challenges associated with microalgal biomass characterization
and the comparison of microalgal strains and conversion processes
is the rapid determination of the composition of algae. We have developed
and applied a high-throughput screening technology based on near-infrared
(NIR) spectroscopy for the rapid and accurate determination of algal
biomass composition. We show that NIR spectroscopy can accurately
predict the full composition using multivariate linear regression
analysis of varying lipid, protein, and carbohydrate content of algal
biomass samples from three strains. We also demonstrate a high quality
of predictions of an independent validation set. A high-throughput
96-well configuration for spectroscopy gives equally good prediction
relative to a ring-cup configuration, and thus, spectra can be obtained
from as little as 10–20 mg of material. We found that lipids
exhibit a dominant, distinct, and unique fingerprint in the NIR spectrum
that allows for the use of single and multiple linear regression of
respective wavelengths for the prediction of the biomass lipid content.
This is not the case for carbohydrate and protein content, and thus,
the use of multivariate statistical modeling approaches remains necessary.