MolTuner
基于团队开发的分子预训练大模型MPG构建的分子性质预测平台和用户自适应的微调平台。
分子性质预测预训练模型
基于团队开发的分子预训练大模型MPG构建的分子性质预测平台和用户自适应的微调平台。提出的PHD策略，在大规模无标注小分子数据集上训练了一个分子图模型MolGNet，可以捕捉有价值的化学见解，从而产生可解释的表示。利用MPG模型，获得了MIT主办的AI药物发现大赛冠军。
参考文献
- [1]Li, P.; Wang, J.; Li, Z.; Qiao, Y.; Liu, X.; Ma, F.; Gao, P.; Song, S.; Xie, G. Pairwise half-graph discrimination: A simple graph-level self-supervised strategy for pre-training graph neural networks. IJCAI 2021.
- [2]Wang, J.; Li, W.; Hou, C.; Tang, X.; Qiao, Y.; Fang, R.; Li, P.; Gao, P.; Xie, G. HCL: Improving Graph Representation with Hierarchical Contrastive Learning. ISWC 2022.
- [3]Li, P.; Li, Y.; Hsieh, C.-Y.; Zhang, S.; Liu, X.; Liu, H.; Song, S.; Yao, X. TrimNet: learning molecular representation from triplet messages for biomedicine. Briefings in Bioinformatics 2021, 22 (4), bbaa266.
- [4]Li, Y.; Li, P.; Yang, X.; Hsieh, C.-Y.; Zhang, S.; Wang, X.; Lu, R.; Liu, H.; Yao, X. Introducing block design in graph neural networks for molecular properties prediction. Chemical Engineering Journal 2021, 414, 128817.
- [5]Li, Y.; Hsieh, C.-Y.; Lu, R.; Gong, X.; Wang, X.; Li, P.; Liu, S.; Tian, Y.; Jiang, D.; Yan, J. An adaptive graph learning method for automated molecular interactions and properties predictions. Nature Machine Intelligence 2022, 4 (7), 645-651.
- [6]Li, P.; Wang, J.; Qiao, Y.; Chen, H.; Yu, Y.; Yao, X.; Gao, P.; Xie, G.; Song, S. An effective self-supervised framework for learning expressive molecular global representations to drug discovery. Briefings in Bioinformatics 2021, 22 (6), bbab109.