*3.2.5 Background subtraction*

Background subtraction is an untargeted data mining technique in which the control and sample datasets are compared, and meticulously subtracted the background noise signals and matrix-related ion signals from the sample datasets. This technique finds the ions that are present in the test sample but not in a control sample. Control sample background subtraction algorithm is developed by Zhang *et al.* for complete removal of the matrix-related signals from the LC–MS/MS analyte dataset and isolation of the metabolite ions of interest. This algorithm is successfully applied for the identification of glutathione (GSH)-reactive metabolites [101].

Background signals from biological matrices and electrical noises were not efficiently removed by the background subtraction alone. To improve the efficiency, Zhu *et al.* developed a retention-time-shift-tolerant background subtraction and noise reduction algorithm (BgS-NoRA) for biological matrices [102]. The addition of noise reduction algorithm to background subtraction algorithm helps in the reduction of unwanted background signals (matrix-related ions) as well as electrical noises in biological matrices.

The limitation of this technique is a requirement of a good control sample containing all the possible matrix signals, and the consistency of run to run chromatographic retention time [101, 103, 104].

## *3.2.6 Control sample comparison*

Control sample comparison is also an untargeted data mining technique in which control is compared with the sample. Metabolite ion chromatographic peaks are checked for their absence in the control sample. This process is tedious and challenging as the drug related metabolites are identified by comparing each metabolite ion in the spectrum of the analyte sample to that of the control sample. This technique is suitable for the identification of all types of metabolites but compared with the background subtraction, this is a less sensitive and selective technique [71, 91, 105].
