Two-hundred and thirty isolates from 10 different genera (Aspergillus, Emericella, Fusarium, Geosmithia, Neosartorya, Penicillium, Pseudallescheria, Scedosporium, Talaromyces, Fomitopsis), investigated during routine diagnostic efforts, were correlated to 22 laboratory-adapted reference MALDI-TOF MS 鈥減roteomic phenotypes鈥? A growth time-course at 30 掳C on Sabouraud agar medium was performed for the 22 鈥減henotypes鈥?at 48, 72, 96 and 120 h points. The best peptide extraction conditions for full recovery of conidia- or asci-producing multihyphal morph structures and the highest intra- and inter-class profiling correlation were identified for the 120 h point spectra dataset, from which an engineered library derived (pre-analytical phase). Fingerprinting classifiers, selected by Wilcoxon/Kruskal-Wallis algorithm, were computed by Genetic Algorithm, Support Vector Machine, Supervised Neuronal Network and Quick Classifier model construction. MS identification (ID) of clinical isolates was referred to genotyping (GT) and, retrospectively, compared to routine morphotyping (MT) IDs (analytical phase).
Proteomic phenotyping is revolutionizing diagnostic mycology as fully reflecting species/morph varieties but often overcoming taxonomic hindrance.