Synthetic Cannabinoid Receptor Agonists Detection Using Fluorescence Spectral Fingerprinting.

Synthetic cannabinoid receptor agonists (SCRAs), termed "Spice" or "K2", are molecules that emulate the effects of the active ingredient of marijuana, and they have gained enormous popularity over the past decade. SCRAs are Schedule 1 drugs that are highly prevalent in the U.K. prison system and among homeless populations. SCRAs are highly potent and addictive. With no way to determine the dose/amount at the point-of care, they pose severe health risks to users, including psychosis, stroke, epileptic seizures, and they can kill. SCRAs are chemically diverse, with over a hundred compounds used as recreational drugs. The chemical diversity of SCRA structures presents a challenge in developing detection modalities. Typically, GC-MS is used for chemical identification; however, this cannot be in place in most settings where detection is critical, e.g., in hospital Emergency Departments, in custody suites/prisons, or among homeless communities. Ideally, real time, point-of-care identification of SCRAs is desirable to direct the care pathway of overdoses and provide information for informed consent. Herein, we show that fluorescence spectral fingerprinting can be used to identify the likely presence of SCRAs, as well as provide more specific information on structural class and concentration (∼1 µg mL-1). We demonstrate that that fluorescence spectral fingerprints, combined with numerical modeling, can detect both parent and combusted material, and such fingerprinting is also practical for detecting them in oral fluids. Our proof-of-concept study suggests that, with development, the approach could be useful in a range of capacities, notably in harm reduction for users of Spice/K2.

Bayesian automatic relevance determination algorithms for classifying gene expression data.

MOTIVATION: We investigate two new Bayesian classification algorithms incorporating feature selection. These algorithms are applied to the classification of gene expression data derived from cDNA microarrays. RESULTS: We demonstrate the effectiveness of the algorithms on three gene expression datasets for cancer, showing they compare well with alternative kernel-based techniques. By automatically incorporating feature selection, accurate classifiers can be constructed utilizing very few features and with minimal hand-tuning. We argue that the feature selection is meaningful and some of the highlighted genes appear to be medically important.