分子图生成-基于Pytorch+变分图自编码器实现的分子图生成算法-附项目源码-优质项目实战.zip

### Requirements - `Python 3.9` - `PyTorch 1.8` - `PyTorch Geometric 2.0.4` - `numpy 1.22.3` - `networkx 2.7.1` ### Data ***ZINC*** dataset from PyTorch Geometric (PyG) dataset collection is being used here. It contains 12k molecular graphs with 38 heavy atoms. Moreover, for each molecular graph, the node features are the types of heavy atoms and the edge features are the types of bonds between them. Besides, every single node in each of the molecular graphs represents an atom and the edge (link) between any two atoms represents the chemical bonding between them. ### Training, Validation, and Testing The current implementation provides three imperative functions:- - To train a new network : `train()` - To validate the network being trained : `val()` - To test a preTrained network : `test()` - The average loss, area under curve score, and associated average precision for the trained model are printed after every epoch. - All hyperparameters to control training and testing of the model, and generation of the molecular graphs are provided in the given `.py` file. ## Generated *ZINC* molecule graphs `gen_graphs` variable is set to *5* in the current implementation to generate *5* new ***ZINC*** molecule graphs, however, it can be set to any number based on the requirements. ### Generated *ZINC* Molecule 1  ### Generated *ZINC* Molecule 2  ### Generated *ZINC* Molecule 3  ### Generated *ZINC* Molecule 4  ### Generated *ZINC* Molecule 5