Fungal melanin contributes to the survival and virulence of pathogenic fungi, such as Fonsecaea pedrosoi, which is responsible for causing chromoblastomycosis. The objective of this study was to employ Fourier transform infrared spectroscopy (FTIR) to predict the melanin content of F. pedrosoi. The melanin content, in percentage, was previously determined using gravimetry for twenty-six clinical isolates. Quintuplicate spectra of each isolate were obtained using attenuated total reflection (ATR) within the range of 4000 to 650 cm-1. To predict the melanin content, modeling was performed using partial least squares regression (PLS) in the region 1800 - 750 cm-1. Two models were tested: PLS and successive projections algorithms for interval selection in partial least squares (iSPA-PLS). The best modeling results were achieved using iSPA-PLS with one factor. The calibration set exhibited a determination coefficient (R2) of 0.9745 and a root mean square error of cross-validation (RMSECV) of 0.0977. In the prediction set, the R2 value was 0.9711, and the root mean square error of prediction (RMSEP) was 0.0999. Modeling with FTIR and multivariate calibration provides a valuable means of predicting fungal melanin content, which is simpler and more robust, thereby contributing to the advancement of this field of study.
Keywords: FTIR spectroscopy; Fonsecaea pedrosoi; Fungal melanin; Partial least squares regression.
Copyright © 2024 Elsevier B.V. All rights reserved.