Deep learning-based molecular generation has extensive applications in many fields, particularly drug discovery. However, majority of current deep generative models (DGMs) are ligand-based and do not consider chemical knowledge in molecu- lar generation process, often resulting in a relatively low success rate. We herein propose a structure-based molecular generative framework with chemical knowl- edge explicitly considered (named PocketFlow), which generates novel ligand mole- cules inside protein binding pockets. In various computational evaluations, Pocket- Flow showed a state-of-the-art performance with generated molecules being 100% chemically valid and highly drug-like. Ablation experiments prove a critical role of chemical knowledge in ensuring the validity and drug-likeness of the generated mol- ecules. We applied PocketFlow to two new target proteins that are related to epigene- tic regulation, HAT1 and YTHDC1, and successfully obtained wet-lab validated bioac- tive compounds. The binding modes of the active compounds with target proteins are close to those predicted by molecular docking, and further confirmed by the X-ray crystal structure. All the results suggest that PocketFlow is a useful deep generative model, capable of generating innovative bioactive molecules from scratch given a protein binding pocket.