Cross-Platform Identification and Classification of Fentanyl Analogs by Tandem Mass Spectrometry (MS/MS)

The abuse of fentanyl and its analogs has caused a large number of deaths worldwide. Due to the rapid development of new fentanyl analogs, the standard mass spectral library is generally incomplete. The spectra obtained by different mass spectrometers by electrospray ionization (ESI) cannot be used across platforms, resulting in the hysteretic detection of novel fentanyl analogs. This paper reports a machine learning classification model based upon a hybrid similarity search. For the purpose of the identification of fentanyl analogs with different mass spectrometers, a mass spectral library based on the conserved fragmentation behavior (CFB) of fentanyl analogs was established. The results show that the identification accuracy of fentanyl analogs by two mass spectrometers is 100%, and the accuracy of classification model is 97.85%. Furthermore, the model may be applied to linear ion trap mass spectrometry (LIT-MS) and quadrupole time-of-flight mass spectrometry (Q-TOF-MS) with classification accuracies of 100 and 98.17%, respectively. This study provides promising technical support for the real-time and cloud computing detection of unknown fentanyl analogs.