Computer assisted determination of ATP in environmental and food samples by bioluminescent assay: comparison of algorithms.

When the bioluminescent assay (BLA) for the determination of adenosine triphosphate (ATP) in environmental and food samples is performed by conventional photometers, a computing procedure is necessary to exclude the mixing time of reagents from the global time of the reaction because it produces a poor analytical precision [M. Mecozzi, A. Amici, G. Visco, Fres. J. Anal. Chem. 357 (1997) 747-751]. In this paper three automated procedures for correcting this interference were tested and compared with illustrate their advantages and drawbacks. The three techniques were the robust exponential regression and the outliers detection method (RER), the smoothing technique of Savitzky-Golay (SSG) and the smoothing technique of Kalman Filter (SKF). RER and SSG were found to be more effective than SKF in minimizing the effect of noise produced by the mixing of reagents. In addition, the application of computing procedures such as RER and SSG allow to improve the detection limit in the BLA performed by conventional photometers.