 This paper proposes a machine learning framework to predict retention times of enantiomers in high-performance liquid chromatography, HPLC. It uses a documentary dataset of chiral molecule retention times to train a graph neural network, GNN, which can then be used to predict the retention times of unknown compounds. This approach has been tested on a variety of chiral molecules and shown to have good accuracy. Furthermore, it can also be used to predict the retention times of multiple columns simultaneously, thus making it more efficient than traditional methods. This article was authored by Haoshu, Jinglong Lin, Dongxiao Zhong, and others.