Enzyme specificity prediction utilizing cross consideration graph neural networks

Enzymes are the molecular machines of life, and a key property that governs their perform is substrate specificity—the power of an enzyme to acknowledge and selectively act on explicit substrates. This specificity originates from the three-dimensional (3D) construction of the enzyme lively website and complex transition state of the response^1,2. Many enzymes can promiscuously catalyze reactions or act on substrates past these for which they had been initially advanced^1,3-5. However, thousands and thousands of recognized enzymes nonetheless lack dependable substrate specificity info, impeding their sensible purposes and complete understanding of the biocatalytic variety in nature. Herein, we developed a cross-attention-empowered SE(3)-equivariant graph neural community structure named EZSpecificity for predicting enzyme substrate specificity, which was educated on a complete tailored database of enzyme-substrate interactions at sequence and structural ranges. EZSpecificity outperformed the prevailing machine studying fashions for enzyme substrate specificity prediction, as demonstrated by each an unknown substrate and enzyme database and 7 proof-of-concept protein households. Experimental validation with eight halogenases and 78 substrates revealed that EZSpecificity achieved a 91.7% accuracy in figuring out the one potential reactive substrate, considerably larger than that of the state-of-the-art mannequin ESP (58.3%). EZSpecificity represents a common machine studying mannequin for correct prediction of substrate specificity for enzymes associated to elementary and utilized analysis in biology and drugs.