Published in Microorganisms | DOI: 10.3390/microorganisms13030501, 24, February 2025
Authors: Simon De Jaegher, Maria D’Aguanno, David Pinzauti and Manuele Biazzo
Overview
The SARS-CoV-2 pandemic has led to an urgent need for effective and rapid diagnostic tools. In the present study, we have evaluated the predictive diagnostic potential of nasal microbiota by analyzing microbial community structures at different taxonomic level resolutions—species, genus, family, order, class and phylum—using Random Forest modelling. A total of 179 nasal swabs from COVID-19-positive (n = 85) and COVID-19-negative (n = 94) individuals were sequenced using a full-length 16S rRNA sequencing (Oxford Nanopore) approach. During each iteration of the Random Forest model, the dataset was randomly split into a training set (70%) and a testing set (30%). Model performance improved with finer taxonomic resolution, achieving the highest accuracy at the Species level (AUROC = 0.821 ± 0.059) and a sensitivity of 55.6% at a specificity threshold of 90%. A progressive decline in AUROC and sensitivity was observed at broader taxonomic levels. Furthermore, Beta diversity analysis supported that microbial community structures are more distinct between COVID-19-positive and COVID-19-negative groups at finer taxonomic levels. These findings highlight the potential role of nasal microbiota profiling as a secondary diagnostic tool for COVID-19, particularly at species- and genus-level classification, and underscore the importance of high taxonomic resolution in microbiome-based diagnostics. However, limited by an uneven sample distribution and the lack of medical evaluations, further large-scale studies are needed before the nasal microbiota can be implemented in the clinical diagnostics of COVID-19.
