App简介

MMPolymer是一个多模态多任务的预训练框架，将一维序列和三维结构信息整合到了聚合物性能预测任务中

最佳实践

1.输入

• PSMILES字符串，例如[*]CC([*])(F)F
• CSV文件，格式可以参考examples文件夹下的exp.csv

2.选择要预测的性质

• Tg：glass transition temperature
• Egc：bandgap (chain)
• Egb：bandgap (bulk)
• Eea：electron affinity
• Ei：ionization energy
• Xc：crystallization tendency
• EPS：dielectric constant
• Nc：refractive index
• Eat：atomization energy

3.提交任务，在outputs文件夹下查看预测结果

参考文献

[1] https://www.polymergenome.org/guide

[2] https://arxiv.org/abs/2406.04727

引用格式

Wang F, Guo W, Cheng M, et al. MMPolymer: A Multimodal Multitask Pretraining Framework for Polymer Property Prediction[C]//Proceedings of the 33nd ACM International Conference on Information and Knowledge Management. 2024.