Brief methodological description. The workflow consists of three primary steps (Figure 2):
  1. DNA extraction and enzymatic digestion, which yield the 2′-deoxyribonucleoside adducts and unmodified 2′-deoxyribonucleosides, follow the established protocol \cite{Gorokhova2020}. A sample (specific tissues or a whole body for small organisms) is homogenized using a method of choice. For DNA extraction, we recommend a fast and easy method using Chelex 100,  an ion exchange resin serving as a chelating agent in binding polyvalent metal ions  \cite{Walsh2013}, which yields a high amount of DNA  \cite{Phillips2012}. Briefly, a suspension of Chelex 100 is added to the sample homogenate and heated to release nucleic acids. After centrifugation, the resulting supernatant is utilized for digestion. Commonly, the enzymatic digestion is applied using nuclease P1 and snake venom phosphodiesterase I, releasing 5'-mono-phosphate nucleosides, which are dephosphorylated by alkaline phosphatase yielding the  nucleosides for analysis, i.e., 2′-deoxyribonucleoside adducts and unmodified 2′-deoxyribonucleosides  \cite{Gorokhova2020}. The digested samples can be stored at −20 °C until analysis by LC-HRMS.    
  2. Liquid chromatography HRMS analysis and data processing, including the bottom-up (targeted) and the top-down (untargeted) approaches. The bottom-up approach targets anticipated DNA adducts from known or expected exposure agents. Consequently, the common drawback of this approach is that adducts from unknown chemical exposure and endogenous processes may not be captured. Therefore, the comprehensive characterization of the adducts induced by known and unknown exogenous exposures and by secondary biological responses leading to adduct formation (endogenous exposure) requires a top-down adductomics approach. Typically, reversed phase chromatography coupled to an Orbitrap HRMS instrumentation in the positive ionization mode can be used for this purpose, employing data-dependent acquisition with an inclusion list or data-independent acquisition; note that MS-based methods for DNA adductomics analysis are reviewed elsewhere \cite{Guo2019}. The resulting high-resolution accurate mass (HRAM) data allows for increased selectivity due to the ability to differentiate the adduct ion signals from isobaric background ions. Screening for DNA adducts takes advantage of the common structural feature of 2′-deoxyribonucleosides, which is a deoxyribose moiety bound to the nucleobase through a glycosidic bond. The product ion spectra of 2′-deoxyribonucleoside adducts include cleavage of the glycosidic bond with a neutral loss of deoxyribose and leading to protonated nucleobase adducts  \cite{Balbo2014b}. The accurate mass measurements further provide sufficient information to determine the molecular formula of the adduct analytes, which are useful in the structural characterization of the adducts and identifying the chemical exposure leading to the adduct formation. The mass of the adduct indicates its structure and identifies the chemical. This strategy identifies the modified nucleobase and the covalently bound chemical but does not identify the modified gene. Recently, open-source databases \cite{La2022,Guo_2020} and software \cite{Walmsley2021,Sousa2021} were developed to support adduct identification and characterization by processing large and complex MS files for non-targeted adduct detection. In particular, an open-source software nLossFinder for peak screening \cite{Sousa2021} and non-targeted DNA adduct detection based on the characteristic neutral loss (deoxyribose moiety, 116.0474 Da) with high mass accuracy (± 5 ppm) and without any prior knowledge of the adducts. To reduce false positives, data filtering is performed using retention time alignment, accurate mass window, MS/MS fragments and any associated metadata available. The output data are normalized peak areas for specific adducts that can be used for the downstream statistical evaluations (Figure 3). More specifically, the measured MS-peak area of each adduct is normalized to that of 2′-deoxyguanosine from the same LC-HRMS run, which circumvents differences in DNA concentration between the individual samples and for any DNA loss during sample preparation. The data format is typically a matrix, with adduct relative abundance and samples (i.e., individuals) given in columns and rows or vice-versa.    
  3. Statistical analysis for the DNA adductome of each sentinel species used for monitoring purposes should include the following steps: (1) screening the output data to exclude non-varying adducts; (2) identification of the background variability; and (3) evaluation of the test samples against the background variability (i.e., reference state). Similar to metabolomics biomarkers, the task is to reduce the high-dimensional data to only a small number of statistically significant features, that are often chemically identified—where each feature corresponds to a mass-to-charge ratio, retention time, and intensity. However, currently, there are no established methods for deriving environmental quality standards from the OMICS data \cite{Henke2023,Machuca-Sepúlveda2023,Ebner2021}; therefore, different approaches should be evaluated and compared using data for different species and environmental settings.  
.