The recent advancements in computational methods and machine learning have significantly propelled various subfields within chemistry and biomedicine. A notable trend is the integration of deep learning with traditional computational techniques to enhance predictive accuracy and efficiency in molecular and protein design. This approach is particularly evident in the development of models for global warming potential (GWP) prediction, molecular structure optimization, and B-cell epitope prediction, where deep learning frameworks are being tailored to specific challenges. Additionally, the use of generative AI for control structure design in chemical processes and the synthesis of ionizable lipids for mRNA delivery systems demonstrate the versatility of these methods. Innovations like Riemannian score matching for molecular structure optimization and the Sequence-Structure-Surface Fitness (S3F) model for protein fitness prediction highlight the potential for multimodal data integration to improve model performance. Furthermore, advancements in 3D molecular structure generation and docking methods underscore the importance of geometric and topological considerations in these models. The introduction of COMET for target prediction and the 3D Interaction Geometric Pre-training for Molecular Relational Learning (3DMRL) further exemplify the growing sophistication in leveraging AI for complex chemical and biological tasks. These developments collectively indicate a shift towards more integrated, scalable, and accurate computational tools that promise to accelerate the discovery and design processes in chemistry and biomedicine.