Decomposition of complex fluorescence spectra containing components with close emission maxima positions and similar quantum yields. Application to fluorescence spectra of proteins.

Despite of widely application of multivariate analysis in chemometrics, problem of resolving closely positioned components in the fluorescence spectra remained unsolved, thus limiting the usage of fluorescence spectroscopy in analytical purpose. In this paper we have described a novel procedure, adapted especially for the analysis of complex fluorescence spectra with multiple, closely positioned components' maxima. The method was first tested on the simulated spectra and then applied on the spectra of proteins whose fluorophores have similar properties of both the excitation and the emission spectra. In this paper, simple but efficient modification of the method was applied. Instead of analyzing full size emission matrix (12 spectra), 9 spectra wide windows were analyzed, and 4 factors (greatest possible number of factors with physical meaning both for actin and simulated spectra) were extracted in each pass. Obtained factor scores were grouped by using the K-means algorithm. Groups of factor scores obtained from K-means algorithm were passed through the one more factor analysis (FA) in order to find one factor that represents each group. Our approach provides resolution of extremely closed spectral components, which is a vital data for protein conformation analysis based on fluorescence spectroscopy.