Partial least squares regression based rapid quantification of intracellular biopolymers from a Sudan black absorption assay.

[Display omitted] • Sudan black absorption, optical density monitoring of intracellular PHA is proposed. • PLS model on autoscaled data can reliably predict polymer content. • Model is resilient to random error perturbation. • Intracellular protein remains the most important predictor in PLS regression. • Method limit of detection for intracellular PHA is quantified. Rapid quantitation of intracellular polyhydroxyalkanoate (PHA) is essential for monitoring and possibly intervening in ongoing bioprocesses. The traditional multistep protocol involving cell separation, lysis, organic solvent extraction, and purification of intracellular polymers is time intensive. In this study, indirect Sudan black (ISB) adsorption data combined with culture optical density and intracellular protein measurements were used to develop partial least squares (PLS) regression models to predict the intracellular PHA concentration in Cupriavidus necator. ISB analysis supported the hypothesized partitioning of Sudan black into residual biomass, PHA, and abiotic compartments. The PLS regression model developed with autoscaled data offered excellent fit on cross-validation (R2 CV of 0.987) and independent prediction (R2 P of 0.994) sets. The method limit of detection for the PLS model (0.036 g l−1) was reasonably low to track biopolymer accumulation in microbial culture, even in early growth phase. The conservative estimate suggests a significantly shorter processing time with the machine learning-based approach (190 min) than with the traditional protocol (940 min) while retaining analytical interchangeability. [ABSTRACT FROM AUTHOR]