Expert Algorithm to Identify Seized Drugs from Tandem Mass Spectra

This proposal addresses several defined research needs of NIJ’s Forensic Science Research and Development Technical Working Group (TWG), including “The ability to identify NPS by comparison to spectra from a different instrument rather than a reference standard” and “Error rate studies on qualitative analysis.” The major goal of this project is to develop the Expert Algorithm for Substance Identification (EASI) to improve the confidence of drug identifications involving tandem mass spectra. Algorithm development will employ a database of replicate spectra of drug standards collected in up to seven different DEA laboratories and the PI’s lab. These instruments will include the use of both Direct Analysis in Real Time (DART)—for rapid screening of samples—and an electrospray ionization source, to demonstrate the algorithm’s compatibility with liquid chromatography-mass spectrometry instruments in future toxicology applications. These inter-laboratory comparisons will include data obtained on two different vendors of DART-MS/MS instruments, thereby demonstrating the universal applicability of the algorithm. The library of test spectra will focus on seized drugs that are known to be especially difficult to resolve from their isomers, including the growing problem of o-, m-, and p-isomers of fluorofentanyl commonly found in DEA’s casework.

    To assist with widespread adoption by practitioners, the PI will collaborate with a major company to incorporate the algorithm into their commercial software. This work will be conducted in parallel with the fundamental algorithm development in the PI’s laboratory. The algorithm is transparent in its design, operation, and decision-making steps, and the algorithm should not be specific to the vendor, instrument, or acquisition conditions. Therefore, the algorithm will not become obsolete as instruments evolve over time but will exist in perpetuity to facilitate crime labs in making rapid drug identifications without running contemporaneous standards. Finally, the error rate of the algorithm will be presented in a variety of probabilistic forms, including estimates of the sensitivity, selectivity, and likelihood ratios. This research will immediately benefit DEA’s operations in the short term and could ultimately benefit both forensic chemists and forensic toxicologists at the national level in the long term. The results will be presented at forensic science conferences, published in peer-reviewed journals, and incorporated into workshops that the PI regularly teaches to forensic practitioners. The database of tandem mass spectra will be made available to the public through an open access data repository. CA/NCF