Significance Identification of enzyme commission (EC) numbers is essential for accurately understanding enzyme functions. Although several EC number prediction tools are available, they have room for further improvement with respect to computation time, precision, coverage, and the total size of the files needed for EC number prediction. Here, we present DeepEC, a deep learning-based computational framework that predicts EC numbers with high precision in a high-throughput manner. DeepEC shows much improved prediction performance when compared with the 5 representative EC number prediction tools that are currently available. DeepEC will be useful in studying enzyme functions by implementing them independently or as part of a third-party software program.