文献调研

CounterResearch:DrugProtAI,SPIDER(http://pmlabstack.pythonanywhere.com/SPIDER),DrugnomeAI  (https://astrazeneca-cgr-publications.github.io/DrugnomeAI/)

DrugProtAI：输入特征：1.Domain（size==20)--节点内信息对应实体 2.Protein-Protein Interactions 3.PPI Properties(size==7)--节点间/边信息对应联系，binary（vivo/vitro) 4. PTM(post-translational modification) Count(size==7):糖基化，交联化，二硫键，信号肽，残基修饰 5.糖基化数目：（O-linked, N-linked,  S-linked, C-linked, O-alpha-linked, and N-beta-linked）6.Subcellular Locations (SCLs) (50 features)7.Flexibility Sequence Properties (FSPs) (14 features) 8.Physicochemical Properties (PCPs) (32 features) 8.Latent Values (20 features) 9.Grouped dipeptide composition (GDPC) Encodings (25 features)