Motivation: Recently, research on independently utilizing either explicit knowledge from knowledge graphs or implicit knowledge from biomedical literature for AI drug discovery has been growing rapidly. These approaches have greatly improved the prediction accuracy of AI models on multiple downstream tasks. However, integrating explicit and implicit knowledge independently hinders their understanding of molecules. Results: We propose DeepEIK, a unified deep learning framework that incorporates both explicit and implicit knowledge for AI drug discovery. We adopt feature fusion to process the multi-modal inputs, and leverage the attention mechanism to denoise the text information. Experiments show that DeepEIK significantly outperforms state-of-the-art methods on crucial tasks in AI drug discovery including drug-target interaction prediction, drug property prediction and protein-protein interaction prediction. Further studies show that benefiting from explicit and implicit knowledge, our framework achieves a deeper understanding of molecules and shows promising potential in facilitating drug discovery applications.