Fig. 2: DDA spectra deconvolution in MetaboAnalystR 4.0.

a The overall workflow. All spectra are evaluated as “clean” or “chimeric” based on the MS signals of the nearest MS1 scan. The reference spectra of all ions (A, B, and C) within the isolation window and above the intensity threshold are extracted from a reference MS2 library for regression analysis. b Diagram of reference spectrum prediction strategy. MetaboAnalystR can predict one or more candidates missing in the MS2 library. Formula(s) of the ion (e.g., ion C) is predicted at first. An abiotic/bio-transformation network is constructed for the formula (e.g., C₂H₇NO₃S), and all neighbors with reference spectra of the formula are extracted as a list. Each fragment of the single spectrum in the list is predicted in a formula. If the formula (e.g., C₃H₉) of the fragment includes more chemical elements (number or type) than the original formula (e.g., C₂H₇NO₃S), it is excluded from the spectrum. The clean spectrum is returned to the list. Their similarities to the original MS2 spectrum of all spectra in the list are evaluated, and the one with the highest similarity score is selected as the predicted spectrum for the ion.