用神经网络法预测mRNA的剪接位点

PREDICTING THE SPLICING SITES OF mRNA BY NEURAL NETWORK

用神经网络预测了mRNA的剪接位点,比较了在各种不同的情况下,神经网络的学习与预测的情况,讨论了能反映真实剪接位点预测情况的有效预测成功率,指出它可达64%,而且总的预测成功率可达98%.预测的相关系数为0.66.

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.