Improvement of drug identification in urine by LC-QqTOF using a probability-based library search algorithm

A common method for identifying an unknown compound involves acquiring its mass spectrum and then comparing that spectrum against a spectral database, or library. Accurate comparison and identification is dependent on the quality of both the library and the test spectrum, but also the search algorithm used. Here, we describe a redesigned probability-based library search algorithm (ProLS) and compare its performance against two predicate algorithms, AMDIS from NIST (NIST) and LibraryView/MasterView (LV/MV), on human urine samples containing drugs of interest that were analyzed by quadrupole-time of flight (QqTOF) mass spectrometry. Each algorithm was used to compare the spectral data collected against an in-house spectral library. ProLS outperformed both NIST and LV/MV in efficiency of drug detection. Additionally, it demonstrated a scoring profile that resulted in an increased likelihood of low match scores for compounds that were absent from a sample. Increased scoring accuracy has the potential to reduce the time that analysts spend manually reviewing match data. Although search algorithms tend to be underappreciated, since they are not typically part of the end-user interface, this work illustrates how a redesigned algorithm can impact the accuracy of identification of small molecules in a biological matrix, and influence the overall utility of a bioanalytical method.