Reducing Inter-Replicate Variation in Fourier Transform Infrared Spectroscopy by Extended Multiplicative Signal Correction

Abstract

Fourier transform infrared (FT-IR) spectroscopy is a powerful tool for characterizing biological tissues and organisms, but it is plagued by replicate variation of various sources. Here, a method for estimating and correcting unwanted replicate variation in multivariate measurement signals, based on extended multiplicative signal correction (EMSC), is presented. Systematic patterns of unwanted methodological variations are estimated from replicate spectra, modeled by a linear subspace model, and implemented into EMSC. The method is applied to FT-IR spectra of two different sets of microorganisms (different double gene knockout strains of Saccharomyces cerevisiae and different species of Listeria) and compared to other preprocessing methods used in FT-IR absorption spectroscopy of microorganisms. The EMSC replicate correction turns out to perform best among the compared methods.