Predicting the biological properties of molecules is crucial in computer-aided drug development, yet it’s often impeded by data scarcity and imbalance in many practical applications. Existing approaches are based on self-supervised learning or 3D data and using an increasing number of parameters to improve performance. These approaches may not take full advantage of established chemical knowledge and could inadvertently introduce noise into the respective model. In this study, we introduce a more elegant transformer-based framework with focused attention for molecular representation (TransFoxMol) to improve the understanding of artificial intelligence (AI) of molecular structure property relationships. TransFoxMol incorporates a multi-scale 2D molecular environment into a graph neural network + Transformer module and uses prior chemical maps to obtain a more focused attention landscape compared to that obtained using existing approaches. Experimental results show that TransFoxMol achieves state-of-the-art performance on MoleculeNet benchmarks and surpasses the performance of baselines that use self-supervised learning or geometry-enhanced strategies on small-scale datasets. Subsequent analyses indicate that TransFoxMol’s predictions are highly interpretable and the clever use of chemical knowledge enables AI to perceive molecules in a simple but rational way, enhancing performance.

Bystander‐killing payloads can significantly overcome the tumor heterogeneity issue and enhance the clinical potential of antibody‐drug conjugates (ADC), but the rational design and identification of effective bystander warheads constrain the broader implementation of this strategy. Here, graph attention networks (GAT) are constructed for a rational bystander killing scoring model and ADC construction workflow for the first time. To generate efficient bystander‐killing payloads, this model is utilized for score‐directed exatecan derivatives design. Among them, Ed9, the most potent payload with satisfactory permeability and bioactivity, is further used to construct ADC. Through linker optimization and conjugation, novel ADCs are constructed that perform excellent anti‐tumor efficacy and bystander‐killing effect in vivo and in vitro. The optimal conjugate T‐VEd9 exhibited therapeutic efficacy superior to DS‐8201 against heterogeneous tumors. These results demonstrate that the effective scoring approach can pave the way for the discovery of novel ADC with promising bystander payloads to combat tumor heterogeneity.