Splice site identification using probabilistic parameters and SVM classification

Abstract

Background: Recent advances and automation in DNA sequencing technology has created a vast amount of DNA sequence data. This increasing growth of sequence data demands better and efficient analysis methods. Identifying genes in this newly accumulated data is an important issue in bioinformatics, and it requires the prediction of the complete gene structure. Accurate identification of splice sites in DNA sequences plays one of the central roles of gene structural prediction in eukaryotes. Effective detection of splice sites requires the knowledge of characteristics, dependencies, and relationship of nucleotides in the splice site surrounding region. A higher-order Markov model is generally regarded as a useful technique for modeling higher-order dependencies. However, their implementation requires estimating a large number of parameters, which is computationally expensive.

High-accuracy splice site prediction based on sequence component and position features

“…Because AUC and auPRC were adopted as the evaluation indices in related published studies Baten et al, 2006Baten et al, , 2008, our results were also translated into those indices for convenience of comparison. Table 4 summarizes the predictive accuracy of our models and other models in terms of the AUC and auPRC for the NN269 dataset.…”

“…For acceptor sites, 208 true samples and 881 pseudo-samples were selected as the test set and the rest as the training set. The selection referred to the references Sonnenburg et al (2007) and Baten et al (2006Baten et al ( , 2008.…”

“…ROC analysis is an effective and widely used method to assess the performance of a classification method (Baten et al, 2006). Plotting the positive predictive value PPV = TP/ (FP + TP), i.e., the fraction of correct positive predictions among all positively predicted examples against Sn, one obtains the precision recall curve (PRC) (Davis and Goadrich, 2006).…”

“…Hence, our method gave the best predictive performance in the dataset NN269. (Durbin et al, 1998); LIK = support vector machine (SVM) using the locality improved kernel (Zien et al, 2000); WD = weighted degree kernel (Rätsch et al, 2004); WDS = weighted degree kernel with shifts (Rätsch et al, 2005); MC-SVM = Markov chain-SVM (Baten et al, 2006); MM1-SVM = first-order Markov model-SVM (Baten et al, 2008); IC-S-SVM = IC Shapiro SVM (Baten et al, 2008); AUC = area under the ROC curve; auPCR = area under the precision recall curve. …”

“…The weight matrix method (WMM) is the earliest and most influential one that uses the position-specific compositional biases (Staden, 1984;Tavares et al, 2009). Subsequently, the pattern recognition algorithms, represented by a Bayesian network (Cai et al, 2000), support vector machine (SVM) (Zhang et al, 2006;Baten et al, 2006;Sonnenburg et al, 2007;Asa et al, 2008;Zhang et al, 2009), hidden Markov model (Baten et al, 2007(Baten et al, , 2008Asa et al, 2008;Zhang et al, 2009Zhang et al, , 2010, artificial neural network (Reese et al, 1997;Wang et al, 2009), etc., were successively introduced. A series of special prediction tools were also improved for splice site prediction, such as GeneSplicer (Pertea et al, 2001), DGSplicer (Chen et al, 2005), NNSplice (Reese et al, 1997), SpliceMachine (Kahn et al, 2007), etc.…”