A pytorch implement of Drug3D-Net1[code] with QM9 2,3 dataset.
And design my network to adapt the voxel input.
master branch - stable version
lab branch - implement version on lab server
- TODO: load data from qm9 dataset and convert into voxel
- TODO: implement the network structure with pytorch
- TODO: rewrite the output of the network, train one property at one time
- TODO: adapt the epoch and batch size for a better performance
- TODO: add Swin Transformer4 in the network
- TODO: design my network to rationally combine the two attention modules
[1]Li C, Wang J, Niu Z, et al. A spatial-temporal gated attention module for molecular property prediction based on molecular geometry[J]. Briefings in Bioinformatics, 2021.[pdf]
[2]Ruddigkeit L, Van Deursen R, Blum L C, et al. Enumeration of 166 billion organic small molecules in the chemical universe database GDB-17[J]. Journal of chemical information and modeling, 2012, 52(11): 2864-2875.[pdf]
[3]Ramakrishnan R, Dral P O, Rupp M, et al. Quantum chemistry structures and properties of 134 kilo molecules[J]. Scientific data, 2014, 1(1): 1-7.[pdf]
[4]Liu Z, Lin Y, Cao Y, et al. Swin transformer: Hierarchical vision transformer using shifted windows[J]. arXiv preprint arXiv:2103.14030, 2021.[pdf]