SeqVec-GAT: A Golgi Classification Model Based on Multi-headed Graph Attention Network

Golgi apparatus is also known as Golgi complex and Golgi apparatus. It is one of the components of the endosomal system in eukaryotic cells. The main function of the Golgi apparatus is to process, sort, and transport proteins synthesized by the endoplasmic reticulum, and then sort them into specific parts of the cell or secrete them outside the cell. Dysregulation of the Golgi apparatus can cause neurodegenerative diseases. The classification of Golgi proteins is particularly important for the development of drugs to treat the corresponding diseases, but existing methods are cost time and laborious. In this paper, we utilize the SeqVec model to extract the features of Golgi proteins and utilize a multi-headed graph attention network as a classification model. The experimental results show that the predictive classification is better than the machine learning methods commonly used in golgi protein classification, the final experimental results are Acc 98.44%, F1-score 0.9844, Sn 92.31%, Sp 100%, MCC 0.9515, AUROC 0.9615.