No abstract

No abstract

34 This study proposes a comparative investigation of different linear and non-linear 35 chemometrics methods applied to the same database of infrared spectra for filamentous fungi 36 discrimination and identification. The database was comprised of 277 strains, (14 genus, 36 37 species), identified and validated by DNA sequencing, and analyzed by high-throughput 38 Fourier Transform Infrared (FTIR) spectroscopy in the 4000-400 cm -1 wavenumber range. A 39 cascade of 20 supervised models based on taxonomic ranks was constructed to predict classes 40 until the species taxonomic rank. The cascade modeling was used to test 11 algorithms (5 41 linear and 6 non-linear) of supervised classification methods. To assess these algorithms, 42 indicators of classification rates and McNemar's tests were defined and applied in same way 43 to each of them. For non-linear algorithms, the KNN (K Nearest Neighbors) method proved to 44 be the best classifier (78%). Linear algorithms, PLS-DA (Partial Least Square -Discriminant 45Analysis) and SVM (Support Vector Machine) showed better performances than non-linear 46 methods with the best classification potential (~93%). SVM and PLS-DA were comparable 47 and a possible complementarity between these two algorithms was highlighted. 48 49