Abstract: Drug-target interaction (DTI) prediction is crucial for identifying newtherapeutics and detecting mechanisms of action. While structure-based methodsaccurately model physical interactions between a drug and its protein target,cell-based assays such as Cell Painting can better capture complex DTIinteractions. This paper introduces MOTI$\mathcal{V}\mathcal{E}$, a MorphologicalcOmpound Target Interaction Graph datasetthat comprises Cell Painting features for $11,000$ genes and $3,600$ compoundsalong with their relationships extracted from seven publicly availabledatabases. We provide random, cold-source (new drugs), and cold-target (newgenes) data splits to enable rigorous evaluation under realistic use cases. Ourbenchmark results show that graph neural networks that use Cell Paintingfeatures consistently outperform those that learn from graph structure alone,feature-based models, and topological heuristics. MOTI$\mathcal{V}\mathcal{E}$accelerates both graph ML research and drug discovery by promoting thedevelopment of more reliable DTI prediction models. MOTI$\mathcal{V}\mathcal{E}$ resources areavailable at https://github.com/carpenter-singh-lab/motive.