PREDICTING THE SPLICING SITES OF mRNA BY NEURAL NETWORK
用神经网络法预测mRNA的剪接位点

We use the neural network to predict the splicing sites of the pre-mRNA, that is to recognize the boundaries of intron-exon and exon-intron from the RNA sequence, then distinguish between e.xons and introns. We analyse the various conditions of learning and prediction by the network, discuss the effictive success rate of real spicing sites, and point out that it is about 64%. The overall success rate of our prediction can reach 98%. The correlation coefficient of the prediction is about 0.66.