The effect of different convolutional kernel sizes on the DeepSTF model's average ACC, ROC-AUC, and PR-AUC. The value k denotes the number of convolutional kernels, the evaluation metric is represented by the x-axis, and the score is represented by the y-axis. Purple represents the performance of the 32 convolutional kernel DeepSTF model, blue represents the performance of the 64 convolutional kernel DeepSTF model, and green represents the performance of the 128 convolutional kernel DeepSTF model.