Computational Removal of Undesired Mass Spectral Features Possessing Repeat Units via a Kendrick Mass Filter

Abstract

Polymers are a common component of chemical background which complicates data analysis and can impair interpretation. Undesired chemical background cannot always be addressed via preanalytical methods, chromatography, or existing data processing methods. The Kendrick Mass Filter (KMF) is presented for the computational removal of undesired signals present in MS 1 spectra. The KMF is analogous to mass defect filtering but utilizes homology information via Kendrick mass scaling in combination with chromatographic retention time and the number of observed signals. The KMF is intended to assist in situations in which current data processing methods to remove background, e.g. blank subtraction, are either not possible or effective. The major parameters affecting KMF were investigated using PEG 400 and NIST standard reference material 1950 (metabolites in human plasma). Further exploration of the KMF performance was tested using an extract of a swab known to contain polymers. An illustrative real-world example of skin analysis with polymeric signal is discussed. The KMF is also able to provide a high-level view of the compositionality of data regarding the presence of signals with repeat units, and indicate the presence of different polymers.

Untargeted metabolomic study of autism in newborn screening samples: a pilot study

“…Using a Kendrick Mass Filter, we explored the compositionality of our data to assess the potential presence of undesired chemical background [104]. Out of the 4,360 features obtained through MZmine preprocessing, more than 1,100 possessed repeat units typical of polyethylene glycol (PEG) and polypropylene glycol (PPG).…”

“…Out of the 4,360 features obtained through MZmine preprocessing, more than 1,100 possessed repeat units typical of polyethylene glycol (PEG) and polypropylene glycol (PPG). Filtering of PEG followed by filtering of PPG was performed using a Kendrick Mass Filter [104] with the following parameters: number of observed signals = 5, Kendrick mass defect = 0.01, and fraction base = 1. All Jupyter notebooks used are publicly available on GitHub (https://github.com/SSI-Metabolomics/Autism_SupplementaryMaterial).…”

“…(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. PEG (4a) and PPG (4b) filtering scripts using the Kendrick Mass filter according to [80] See the two .ipynb…”

“…Using a Kendrick Mass Filter, we explored the compositionality of our data to assess the potential presence of undesired chemical background [104]. Of the 3,253 remaining features, we further excluded those with a maximum intensity in paper blanks / maximum intensity in samples ratio ≥ 0.2, as well as features with 80% or more gaps (i.e.…”

Metaproteomic and Metabolomic Approaches for Characterizing the Gut Microbiome

“…Although DDA approaches are able to provide high quality MS/MS spectra with minimal interferences, limitations including incomplete spectra databases, limited tandem mass spectra coverage, lack of FDR estimation for spectrum matching scores and the presence of interfering ions that restrict DDA‐based metabolite identifications . Methods have been developed to resolve these issues, such as molecular networking, Kendrick mass filtering and FDR estimation for spectra matching, and are reviewed elsewhere . In contrast, DIA approaches, which typically include all‐ion fragmentation (AIF) and sequential window acquisition of all the theoretical fragment‐ion spectra (SWATH), generate MS/MS spectra of all precursor ions simultaneously within selection window.…”