Recently, scientists published a study on protein-ligand complex structure prediction in the journal 'Nature Machine Intelligence'. The new method, NeuralPLexer, predicts structures using deep generative models with only the protein sequence and ligand molecule graph as input. NeuralPLexer leverages a multiscale deep learning system to enhance prediction success rates and is expected to play a significant role in drug discovery. This research provides insights for exploring the field of protein-ligand complex structure prediction.