中文摘要：

基于多个软件的预测结果，提出了一种融合方法（ComPromoter）提高真核生物基因启动子的预测精度．ComPromoter以ProKey、FirstEF和Eponine等软件对启动子的预测结果为基础，融合了可用于启动子识别的多个特征（包括ProKey预测结果中真阳性和假阳性的出现频率随投票距离以及预测分值变化的统计特征），并利用支持向量机构建了启动子预测模型．在人类ENCODE区域测试数据集上测试的结果显示，ComPromoter的Pearson相关系数CC值高于所融合的单个启动子预测软件．

英文摘要：

A new method, named ComPromoter, is proposed to combine several promoter predictors for improving the prediction accuracy of eukaryotic promoter. Based on the outputs of three promoter predictors including ProKey, FirstEF and Eponine, ComPromoter integrated several features, including the relationship between the occurrence frequencies of outputs （including true positives and false positives） of ProKey and the voting distances, and the relationship between the occurrence frequencies of outputs of ProKey and prediction scores. Testing results on the human ENCODE test regions showed that ComProrooter could achieve higher Pearson correlation coefficient （CC） than any combined promoter predictor.