【摘要】 目的:改进转录因子结合位点的理论预测方法。方法:构建转录因子结合位点位置权重矩阵,以转录因子结合位点每一位置的碱基保守性指数Mi为参量,利用位置权重打分函数算法(PWMSA)对酵母五种转录因子结合位点进行预测。结果:利用self-consistency和cross-validation两种方法对此算法进行检验,均获得了较高的预测成功率,结果表明5种转录因子结合位点的预测成功率均超过80%。结论:与已有的三种预测转录因子结合位点的软件进行比较,PWMSA算法明显优于其他三种算法,核苷酸水平上的关联系数和结合位点水平上的关联系数分别提高了0.25和0.22。更多还原

【Abstract】 Objective:To improve the predictive capacity of the algorithm for transcription factor binding sites.Methods: By constructing position weight matrix and calculating the site conservative index vectors Mi in transcription factor binding sites,a novel position weight matrix scoring algorithm(PWMSA) for predicting yeast transcription factor binding sites is presented.Results: By using of the self-consistency test and the 10-fold cross-validation test,prediction results show that the more than 80% of correct prediction are obtained for all of five transcription factor binding sites.Conclusion: By comparing our algorithm with other three softwares,the results show that the overall prediction accuracies of PWMSA are more 0.22 and 0.25 than the other three algorithms,respectively,at binding sites segment level and nucleotide level.更多还原