Exploring the Complexities of Dissolved Organic Matter Photochemistry from the Molecular Level by Using Machine Learning Approaches

Abstract

Dissolved organic matter (DOM) sustains a substantial part of the organic matter transported seaward, where photochemical reactions significantly affect its transformation and fate. The irradiation experiments can provide valuable information on the photochemical reactivity (photolabile, photoresistant, and photoproduct) of molecules. However, the inconsistency of the fate of irradiated molecules among different experiments curtailed our understanding of the roles the photochemical reactions have played, which cannot be properly addressed by traditional approaches. Here, we conducted irradiation experiments for samples from two large estuaries in China. Molecules that occurred in irradiation experiments were characterized by the Fourier transform ion cyclotron resonance mass spectrometry and assigned probabilistic labels to define their photochemical reactivity. These molecules with probabilistic labels were used to construct a learning database for establishing a suitable machine learning (ML) model. We further applied our well-trained ML model to "unmatched" (i.e., not detected in our irradiation experiments) molecules from five estuaries worldwide, to predict their photochemical reactivity. Results showed that numerous molecules with strong photolability can be captured solely by the ML model. Moreover, comparing DOM photochemical reactivity in five estuaries revealed that the riverine DOM chemistry largely determines their subsequent photochemical transformation. We offer an expandable and renewable approach based on ML to compatibly integrate existing irradiation experiments and shed insight into DOM transformation and degradation processes.

Machine Learning Models for Evaluating Biological Reactivity Within Molecular Fingerprints of Dissolved Organic Matter Over Time

“…Alternatively, incubations could be dedicatedly developed corresponding to the scenario in the field to simulate the biogeochemical processes over different time scales. Therefore, time‐specific incubation combined with simple ML tools represents a feasible and effective method to predict the reactivity of MFs in field samples that cannot be captured by limited incubations (Zhao et al., 2023). We find that longer incubation designed for longer water retention time yields better model performance.…”

“…As an emerging artificial intelligence domain, ML can learn from large, complex, and multi-dimensional data to develop predictive models and has proven to be a promising tool for handling non-linear correlations between the learnable features and target labels (Bergen et al, 2019). As such, ML models have been established and well applied to FT-ICR MS data analysis to gain an in-depth understanding of both the bulk (Yi et al, 2023) and molecular level (Plamper et al, 2023;Zhao et al, 2023). In previous work, we constructed ML models to study the photochemical reactivity of estuarine DOM worldwide (Zhao et al, 2023).…”

“…As such, ML models have been established and well applied to FT-ICR MS data analysis to gain an in-depth understanding of both the bulk (Yi et al, 2023) and molecular level (Plamper et al, 2023;Zhao et al, 2023). In previous work, we constructed ML models to study the photochemical reactivity of estuarine DOM worldwide (Zhao et al, 2023). However, the model did not account for the effects of incubations with various durations.…”

Reassessing the Quantum Yield and Reactivity of Triplet-State Dissolved Organic Matter via Global Kinetic Modeling

“…Dissolved organic matter (DOM) is a heterogeneous mixture containing complex molecules that are ubiquitous in aquatic environments. , DOM can absorb sunlight and generate photochemically produced reactive intermediates (PPRIs) including excited triplet-state DOM ( 3 DOM*), singlet oxygen ( 1 O 2 ), hydroxyl radical ( • OH), and others. − Notably, 3 DOM* is both the major source of other PPRIs and a strong oxidant itself; thus, it has significant impacts on the transformation of aquatic contaminants, biomolecules, and element cycling. ,− However, unlike other PPRIs that exist as single and distinct species, 3 DOM* is a complex assortment of excited states at various triplet energies and reduction potentials. , Consequently, it is challenging to accurately characterize its properties, − and most available measurement techniques can only provide generalized information. ,, …”

“…However, to discern noticeable kinetic variations across different [TMP] 0 , researchers often employ a wide [TMP] 0 range, typically from 50 to 800 μM. ,,, Given this range, the steady-state assumption may not always hold, possibly leading to inaccuracies in the estimation of R 3 DOM* and k 3 DOM*,TMP . Overall, while current methods for studying 3 DOM* kinetics are effective under specific conditions, further refinement could improve modeling accuracy and interpretability, thereby enhancing the reliability of subsequent studies. ,, …”