With the fast development of artificial intelligence, a lot of translation methods and search methods have been proposed to address molecular optimization problems in drug design, which enables this field to achieve remarkable progress. However, existing methods still encounter great difficulties in addressing problems involving more than three properties, since these problems pose stiff challenges to translation methods and search methods in terms of acquiring high-quality training data and balancing multiple properties, respectively. In this paper, we propose an adaptive evolutionary optimization framework to address the many-property molecular optimization problems (namely MaOMO). MaOMO adaptively identifies the property with the largest improvement potential in each iteration, which generates high-quality molecules as efficiently as possible by devoting more efforts to the property. Besides, MaOMO adopts a dynamic selection strategy to select molecules with large property improvement, good property diversity, and structure diversity. We investigate the performance of MaOMO framework on both benchmark and practical molecular optimization tasks, which involve the simultaneous optimization of four or more properties. Experimental results show that the proposed framework is superior to five state-of-the-art competitors, which achieves a success rate improvement of more than 20% on practical optimization tasks.

In the field of cancer treatment, drug combination therapy appears to be a promising treatment strategy compared to monotherapy. Recently, plenty of computational models are gradually applied to prioritize synergistic drug combinations. However, the existing prediction models have not fully exploited the multi-way relations between drug combinations and cell lines. Besides, the number of identified drug-drug-cell line triplets is insufficient owning to the high cost of in vitro screening, which affects the ability of models to capture and utilize multi-way relations. To address this challenge, we design the multi-view hypergraph contrastive learning model, termed MHCLSyn, for synergistic drug combination prediction. First, the synergistic drug-drug-cell line triplets are formulated as a drug synergy hypergraph, and three task-specific hypergraphs are designed based on the drug synergy hypergraph. Then, we design a multi-view hypergraph contrastive learning with enhancement schemes, which allows for more expressive and discriminative node representation learning on drug synergy hypergraph. After that, the representations of nodes indicating drug-drug-cell line triplets are inputted to fully connected network for making predictions. Extensive experiments show MHCLSyn achieves better performance than state-of-the-art prediction models on benchmark datasets and is applicable to unseen drug combinations or cell lines. Case study indicates that MHCLSyn is capable of detecting potential synergistic drug combinations.