微孢子虫PolyA位点的预测

多聚腺苷酸化是真核细胞内形成成熟mRNA的一个重要步骤,其位点的预测对基因组序列中编码基因的发掘具有重要的参考价值.本研究以缺乏有效基因预测方法的微孢子虫基因组为对象,根据该物种的基因表达偏好设计了一个算法,对其PolyA位点进行预测分析.首先,采用k阶核苷酸频率形式和位置权重矩阵形成初始的特征,然后用PCA降低特征空间的维数,得到的数据用机器学习方法进行分析,产生一个较好的分类结果.其中基于支持向量机的实验得到的敏感度(Sp)和ACC分别达到了87.33%和85.14%,这在微孢子虫的PolyA位点预测上取得了较为理想的效果,并为以后机器学习算法在微孢子虫基因预测领域做了很好的尝试.

Prediction of Polyadenylation Sites in Microsporidian Genome

Polyadenylation is a critical cellular process that forms mature mRNAs in eukaryotic cells. The prediction of its sites is of an important reference value for the discovery of encoding genes in the genome sequence. At present,no effective gene prediction methods for microsporidian genomes are available. Here,we studied microsporidia genomes and,according to the preference of gene expression of the species,proposed a method to predict and analyze poly(A) sites of microsporidium. First,we employed the K-gram nucleotide acid pattern,position weight matrix and increment of diversity to form the initial features. Then we used PCA to reduce the dimension of the initial feature space. Finally,a classification model integrating SVM classifiers was built to predict poly(A) sites. By the proposed algorithm,we achieved a specificity (Sp) of 87.33% and an accuracy (ACC) of 85.14% in the specific dataset. This method also gave an ideal result in the prediction of the poly(A) sites in the microsporidium genome.