Location: Quality & Safety Assessment ResearchTitle: Salmonella detection from chicken rinsate with surface enhanced Raman spectroscopy and RT-PCR validation
|SETIA, GAYATRI - Stoneybrook University|
Submitted to: Poultry International Exposition
Publication Type: Abstract Only
Publication Acceptance Date: 11/30/2015
Publication Date: 1/25/2016
Citation: Eady, M.B., Park, B., Setia, G. 2016. Salmonella detection from chicken rinsate with surface enhanced Raman spectroscopy and RT-PCR validation. Poultry International Exposition. P208.
Technical Abstract: Optical detection of bacteria has been approached in recent years as a bacteria detection method that can counter time restraints of traditional plating or the high reoccurring cost of real-time polymerase chain reaction (RT-PCR). The goal of optical detection is to identify bacteria with spectral signatures unique to the organisms. The objective of this research is to develop a detection method with surface enhanced Raman spectroscopy (SERS) to classify bacterial colonies recovered from chicken rinsates as either Salmonella positive or negative, while comparing results to RT-PCR. The SERS spectra were enhanced by the nanosubstrate consisting of biopolymer encapsulated AgNO3. Bacterial colonies from chicken rinsates were grown on Brilliant green with sulfapyridine (BGS) agar. Individual colonies (n = 69) were sampled by PCR, while preparing bacterial suspensions to air dry onto slides with a dried layer of the nanoparticle substrate. Particle size data were collected over six months to assess the stability of the nanoparticle substrate, and found a mean size of 43.32 nm with a mean variation of ± 2.67 nm over 14 measurements. Ten spectra from each sample were collected between 301 – 1800 cm-1. Spectral peaks from the SERS Rayleigh light scattering of both Salmonella positive (n = 410) and negative samples (n = 280) were identified, and then labeled as corresponding cellular components. T-tests were used to determine if peaks were significantly different between positive and negative samples, with five peaks determined to be significantly different at p < 0.0045, with Bonferroni’s correction factor. A support vector machine (SVM) algorithm was applied to classify the samples as either Salmonella positive or negative. The algorithm used a radial-basis function kernel, with classification accuracies of 98.05% for Salmonella positives and 94.64% for Salmonella negatives, using the RT-PCR results as the true classification. It was determined that structural components of tyrosine (692 cm-1), glycosidic ring/adenine/CH2 rocking (718 cm-1), a previously observed, but unidentified peak (791 cm-1), structural components of membrane phospholipids (859 cm-1), and lipid components from the cell wall (1,018 cm-1) were impacting the classification of samples.