Biological carbon fixation benefits evaluation model construction and application based on atomic economy concept

Current methods for evaluating biocarbon fixation efficiency (BCF), such as genome-scale metabolic models and life cycle assessment, lack consideration of reaction atom economy and fail to connect genetic relationships with the reaction process. To address these limitations, we introduced an atomic economy evaluation index centered on enzyme kinetics, named Economic Indicators of Real Biological Carbon Fixation Atoms (EIRCBFA), and proposed a machine learning-based model to assess BCF at both the reaction conditions and protein levels. Using gradient boosting, the models achieved R ² values of 0.853 and 0.937, respectively, in five-fold cross-validation. The model was validated by optimizing dihydroxyacetone (DHA) biosynthesis, where predictions were consistent with traditional carbon efficiency trends. Notably, the highest EIRCBFA mutant, FLS_F484E, produced 33.19 mg/L DHA, with a yield and carbon efficiency three times that of the wild-type enzyme. RAEKP provides a valuable tool for optimizing biocarbon pathways and evaluating their true biocarbon fixed atom economy.