Patrick Williamson (Class of ’16)
SERS spectra of 12 bacterial strains of urinary tract infection (UTI) clinical isolates grown and enriched from urine are reported. A partial least squares-discriminant analysis (PLS-DA) classification treatment of these SERS spectra results in strain level identification with >95% sensitivity and >99% specificity. The classification model successfully identified the SERS spectra of a urine-cultured strain not used to build this statistical model. Enrichment was accomplished by a filtration and centrifugation protocol. The predetermined drug susceptibility profiles of these clinical isolates thus allowed the SERS methodology to provide appropriate UTI antibiotic information in less than 1 h. Most of this time was used for sample preparation procedures (enrichment and washing) for this proof of principle study. SERS spectra of the enriched bacterial samples are dominated by nucleotide degradation metabolites: adenine, hypoxanthine, xanthine, guanine, uric acid, AMP, and guanosine. Strain-specific specificity is due to the different relative amounts of these purines contributing to the corresponding SERS spectra of these clinical isolates. All measurements were made at the minimal bacterial concentration in urine for UTI diagnosis (105 cfu/mL).