Using amino acid patterns to accurately predict translation initiation sites

The translation initiation site (TIS) prediction problem is about how to correctly identify TIS in mRNA, cDNA, or other types of genomic sequences. High prediction accuracy can be helpful in a better understanding of protein coding from nucleotide sequences. This is an important step in genomic analysis to determine protein coding from nucleotide sequences. In this paper, we present an in silico method to predict translation initiation sites in vertebrate cDNA or mRNA sequences. This method consists of three sequential steps as follows. In the first step, candidate features are generated using k-gram amino acid patterns. In the second step, a small number of top-ranked features are selected by an entropy-based algorithm. In the third step, a classification model is built to recognize true TISs by applying support vector machines or ensembles of decision trees to the selected features. We have tested our method on several independent data sets, including two public ones and our own extracted sequences. The experimental results achieved are better than those reported previously using the same data sets. Our high accuracy not only demonstrates the feasibility of our method, but also indicates that there might be "amino acid" patterns around TIS in cDNA and mRNA sequences.     

Recognition of translation initiation sites of eukaryotic genes based on an EM algorithm.

With the rapid increase of DNA databases of human and other eukaryotic model organisms, a large great number of genes need to be distinguished from the DNA databases. Exact recognition of translation initiation sites (TISs) of eukaryotic genes is very important to understand the translation initiation process, predict the detailed structure of eukaryotic genes, and annotate uncharacterized sequences. The problem has not been solved satisfactorily, especially for recognizing TISs of the eukaryotic genes with shorter first exons. It is an important task for extracting new features and finding new powerful algorithms for recognizing TISs of eukaryotic genes. In this paper, the important characteristics of shorter flanking fragments around TISs are extracted and an expectation-maximization (EM) algorithm based on incomplete data is used to recognize TISs of eukaryotic genes. The accuracy is up to 87.8% over a six-fold cross-validation test. The result shows that the identification variables are effectively extracted and the EM algorithm is a powerful tool to predict the TISs of eukaryotic genes. The algorithm also can be applied to other classification or clustering tasks in bioinformatics.     

Mismatch string kernels for discriminative protein classification

MOTIVATION: Classification of proteins sequences into functional and structural families based on sequence homology is a central problem in computational biology. Discriminative supervised machine learning approaches provide good performance, but simplicity and computational efficiency of training and prediction are also important concerns. RESULTS: We introduce a class of string kernels, called mismatch kernels, for use with support vector machines (SVMs) in a discriminative approach to the problem of protein classification and remote homology detection. These kernels measure sequence similarity based on shared occurrences of fixed-length patterns in the data, allowing for mutations between patterns. Thus, the kernels provide a biologically well-motivated way to compare protein sequences without relying on family-based generative models such as hidden Markov models. We compute the kernels efficiently using a mismatch tree data structure, allowing us to calculate the contributions of all patterns occurring in the data in one pass while traversing the tree. When used with an SVM, the kernels enable fast prediction on test sequences. We report experiments on two benchmark SCOP datasets, where we show that the mismatch kernel used with an SVM classifier performs competitively with state-of-the-art methods for homology detection, particularly when very few training examples are available. Examination of the highest-weighted patterns learned by the SVM classifier recovers biologically important motifs in protein families and superfamilies.     

原核基因翻译起始位点预测的新方法

翻译起始位点（TIS,即基因5’端）的精确定位是原核生物基因预测的一个关键问题,而基因组GC含量和翻译起始机制的多样性是影响当前TIS预测水平的重要因素．结合基因组结构的复杂信息（包括GC含量、TIS邻近序列及上游调控信号、序列编码潜能、操纵子结构等）,发展刻画翻译起始机制的数学统计模型,据此设计TIS预测的新算法MED．StartPlus．并将MED．StartPlus与同类方法RBSfinder、GS．Finder、MED-Start、TiCo和Hon-yaku等进行系统地比较和评价．测试针对两种数据集进行：当前14个已知的TIS被确认的基因数据集,以及300个物种中功能已知的基因数据集．测试结果表明,MED-StartPlus的预测精度在总体上超过同类方法．尤其是对高GC含量基因组以及具有复杂翻译起始机制的基因组,MED-StartPlus具有明显的优势．     

Protein homology detection using string alignment kernels

MOTIVATION: Remote homology detection between protein sequences is a central problem in computational biology. Discriminative methods involving support vector machines (SVMs) are currently the most effective methods for the problem of superfamily recognition in the Structural Classification Of Proteins (SCOP) database. The performance of SVMs depends critically on the kernel function used to quantify the similarity between sequences. RESULTS: We propose new kernels for strings adapted to biological sequences, which we call local alignment kernels. These kernels measure the similarity between two sequences by summing up scores obtained from local alignments with gaps of the sequences. When tested in combination with SVM on their ability to recognize SCOP superfamilies on a benchmark dataset, the new kernels outperform state-of-the-art methods for remote homology detection. AVAILABILITY: Software and data available upon request.     

Prediction of both conserved and nonconserved microRNA targets in animals

MOTIVATION: MicroRNAs (miRNAs) are involved in many diverse biological processes and they may potentially regulate the functions of thousands of genes. However, one major issue in miRNA studies is the lack of bioinformatics programs to accurately predict miRNA targets. Animal miRNAs have limited sequence complementarity to their gene targets, which makes it challenging to build target prediction models with high specificity. RESULTS: Here we present a new miRNA target prediction program based on support vector machines (SVMs) and a large microarray training dataset. By systematically analyzing public microarray data, we have identified statistically significant features that are important to target downregulation. Heterogeneous prediction features have been non-linearly integrated in an SVM machine learning framework for the training of our target prediction model, MirTarget2. About half of the predicted miRNA target sites in human are not conserved in other organisms. Our prediction algorithm has been validated with independent experimental data for its improved performance on predicting a large number of miRNA down-regulated gene targets. AVAILABILITY: All the predicted targets were imported into an online database miRDB, which is freely accessible at http://mirdb.org.     

Global discriminative learning for higher-accuracy computational gene prediction

Most ab initio gene predictors use a probabilistic sequence model, typically a hidden Markov model, to combine separately trained models of genomic signals and content. By combining separate models of relevant genomic features, such gene predictors can exploit small training sets and incomplete annotations, and can be trained fairly efficiently. However, that type of piecewise training does not optimize prediction accuracy and has difficulty in accounting for statistical dependencies among different parts of the gene model. With genomic information being created at an ever-increasing rate, it is worth investigating alternative approaches in which many different types of genomic evidence, with complex statistical dependencies, can be integrated by discriminative learning to maximize annotation accuracy. Among discriminative learning methods, large-margin classifiers have become prominent because of the success of support vector machines (SVM) in many classification tasks. We describe CRAIG, a new program for ab initio gene prediction based on a conditional random field model with semi-Markov structure that is trained with an online large-margin algorithm related to multiclass SVMs. Our experiments on benchmark vertebrate datasets and on regions from the ENCODE project show significant improvements in prediction accuracy over published gene predictors that use intrinsic features only, particularly at the gene level and on genes with long introns.     

Classification of small molecules by two- and three-dimensional decomposition kernels

MOTIVATION: Several kernel-based methods have been recently introduced for the classification of small molecules. Most available kernels on molecules are based on 2D representations obtained from chemical structures, but far less work has focused so far on the definition of effective kernels that can also exploit 3D information. RESULTS: We introduce new ideas for building kernels on small molecules that can effectively use and combine 2D and 3D information. We tested these kernels in conjunction with support vector machines for binary classification on the 60 NCI cancer screening datasets as well as on the NCI HIV data set. Our results show that 3D information leveraged by these kernels can consistently improve prediction accuracy in all datasets. AVAILABILITY: An implementation of the small molecule classifier is available from http://www.dsi.unifi.it/neural/src/3DDK.     

Prediction of protein subcellular locations by support vector machines using compositions of amino acids and amino acid pairs

MOTIVATION: The subcellular location of a protein is closely correlated to its function. Thus, computational prediction of subcellular locations from the amino acid sequence information would help annotation and functional prediction of protein coding genes in complete genomes. We have developed a method based on support vector machines (SVMs). RESULTS: We considered 12 subcellular locations in eukaryotic cells: chloroplast, cytoplasm, cytoskeleton, endoplasmic reticulum, extracellular medium, Golgi apparatus, lysosome, mitochondrion, nucleus, peroxisome, plasma membrane, and vacuole. We constructed a data set of proteins with known locations from the SWISS-PROT database. A set of SVMs was trained to predict the subcellular location of a given protein based on its amino acid, amino acid pair, and gapped amino acid pair compositions. The predictors based on these different compositions were then combined using a voting scheme. Results obtained through 5-fold cross-validation tests showed an improvement in prediction accuracy over the algorithm based on the amino acid composition only. This prediction method is available via the Internet.     

Support vector machines for separation of mixed plant-pathogen EST collections based on codon usage

MOTIVATION: Discovery of host and pathogen genes expressed at the plant-pathogen interface often requires the construction of mixed libraries that contain sequences from both genomes. Sequence identification requires high-throughput and reliable classification of genome origin. When using single-pass cDNA sequences difficulties arise from the short sequence length, the lack of sufficient taxonomically relevant sequence data in public databases and ambiguous sequence homology between plant and pathogen genes. RESULTS: A novel method is described, which is independent of the availability of homologous genes and relies on subtle differences in codon usage between plant and fungal genes. We used support vector machines (SVMs) to identify the probable origin of sequences. SVMs were compared to several other machine learning techniques and to a probabilistic algorithm (PF-IND) for expressed sequence tag (EST) classification also based on codon bias differences. Our software (Eclat) has achieved a classification accuracy of 93.1% on a test set of 3217 EST sequences from Hordeum vulgare and Blumeria graminis, which is a significant improvement compared to PF-IND (prediction accuracy of 81.2% on the same test set). EST sequences with at least 50 nt of coding sequence can be classified using Eclat with high confidence. Eclat allows training of classifiers for any host-pathogen combination for which there are sufficient classified training sequences. AVAILABILITY: Eclat is freely available on the Internet (http://mips.gsf.de/proj/est) or on request as a standalone version. CONTACT: friedel@informatik.uni-muenchen.de.     

Support vector machines for predicting rRNA-, RNA-, and DNA-binding proteins from amino acid sequence

Classification of gene function remains one of the most important and demanding tasks in the post-genome era. Most of the current predictive computer methods rely on comparing features that are essentially linear to the protein sequence. However, features of a protein nonlinear to the sequence may also be predictive to its function. Machine learning methods, for instance the Support Vector Machines (SVMs), are particularly suitable for exploiting such features. In this work we introduce SVM and the pseudo-amino acid composition, a collection of nonlinear features extractable from protein sequence, to the field of protein function prediction. We have developed prototype SVMs for binary classification of rRNA-, RNA-, and DNA-binding proteins. Using a protein's amino acid composition and limited range correlation of hydrophobicity and solvent accessible surface area as input, each of the SVMs predicts whether the protein belongs to one of the three classes. In self-consistency and cross-validation tests, which measures the success of learning and prediction, respectively, the rRNA-binding SVM has consistently achieved >95% accuracy. The RNA- and DNA-binding SVMs demonstrate more diverse accuracy, ranging from approximately 76% to approximately 97%. Analysis of the test results suggests the directions of improving the SVMs.     

Bio-support vector machines for computational proteomics

MOTIVATION: One of the most important issues in computational proteomics is to produce a prediction model for the classification or annotation of biological function of novel protein sequences. In order to improve the prediction accuracy, much attention has been paid to the improvement of the performance of the algorithms used, few is for solving the fundamental issue, namely, amino acid encoding as most existing pattern recognition algorithms are unable to recognize amino acids in protein sequences. Importantly, the most commonly used amino acid encoding method has the flaw that leads to large computational cost and recognition bias. RESULTS: By replacing kernel functions of support vector machines (SVMs) with amino acid similarity measurement matrices, we have modified SVMs, a new type of pattern recognition algorithm for analysing protein sequences, particularly for proteolytic cleavage site prediction. We refer to the modified SVMs as bio-support vector machine. When applied to the prediction of HIV protease cleavage sites, the new method has shown a remarkable advantage in reducing the model complexity and enhancing the model robustness.     

Multi-class protein fold recognition using support vector machines and neural networks

MOTIVATION: Protein fold recognition is an important approach to structure discovery without relying on sequence similarity. We study this approach with new multi-class classification methods and examined many issues important for a practical recognition system. RESULTS: Most current discriminative methods for protein fold prediction use the one-against-others method, which has the well-known 'False Positives' problem. We investigated two new methods: the unique one-against-others and the all-against-all methods. Both improve prediction accuracy by 14-110% on a dataset containing 27 SCOP folds. We used the Support Vector Machine (SVM) and the Neural Network (NN) learning methods as base classifiers. SVMs converges fast and leads to high accuracy. When scores of multiple parameter datasets are combined, majority voting reduces noise and increases recognition accuracy. We examined many issues involved with large number of classes, including dependencies of prediction accuracy on the number of folds and on the number of representatives in a fold. Overall, recognition systems achieve 56% fold prediction accuracy on a protein test dataset, where most of the proteins have below 25% sequence identity with the proteins used in training.     

Subcellular localization prediction with new protein encoding schemes

Subcellular localization is one of the key properties in functional annotation of proteins. Support vector machines (SVMs) have been widely used for automated prediction of subcellular localizations. Existing methods differ in the protein encoding schemes used. In this study, we present two methods for protein encoding to be used for SVM-based subcellular localization prediction: n-peptide compositions with reduced amino acid alphabets for larger values of n and pairwise sequence similarity scores based on whole sequence and N-terminal sequence. We tested the methods on a common benchmarking data set that consists of 2,427 eukaryotic proteins with four localization sites. As a result of 5-fold cross-validation tests, the encoding with n-peptide compositions provided the accuracies of 84.5, 88.9, 66.3, and 94.3 percent for cytoplasmic, extracellular, mitochondrial, and nuclear proteins, where the overall accuracy was 87.1 percent. The second method provided 83.6, 87.7, 87.9, and 90.5 percent accuracies for individual locations and 87.8 percent overall accuracy. A hybrid system, which we called PredLOC, makes a final decision based on the results of the two presented methods which achieved an overall accuracy of 91.3 percent, which is better than the achievements of many of the existing methods. The new system also outperformed the recent methods in the experiments conducted on a new-unique SWISSPROT test set     

Predicting G-protein coupled receptors-G-protein coupling specificity based on autocross-covariance transform

Determining G-protein coupled receptors (GPCRs) coupling specificity is very important for further understanding the functions of receptors. A successful method in this area will benefit both basic research and drug discovery practice. Previously published methods rely on the transmembrane topology prediction at training step, even at prediction step. However, the transmembrane topology predicted by even the best algorithm is not of high accuracy. In this study, we developed a new method, autocross-covariance (ACC) transform based support vector machine (SVM), to predict coupling specificity between GPCRs and G-proteins. The primary amino acid sequences are translated into vectors based on the principal physicochemical properties of the amino acids and the data are transformed into a uniform matrix by applying ACC transform. SVMs for nonpromiscuous coupled GPCRs and promiscuous coupled GPCRs were trained and validated by jackknife test and the results thus obtained are very promising. All classifiers were also evaluated by the test datasets with good performance. Besides the high prediction accuracy, the most important feature of this method is that it does not require any transmembrane topology prediction at either training or prediction step but only the primary sequences of proteins. The results indicate that this relatively simple method is applicable. Academic users can freely download the prediction program at http://www.scucic.net/group/database/Service.asp.     

Metabolic network prediction through pairwise rational kernels

Background

Metabolic networks are represented by the set of metabolic pathways. Metabolic pathways are a series of biochemical reactions, in which the product (output) from one reaction serves as the substrate (input) to another reaction. Many pathways remain incompletely characterized. One of the major challenges of computational biology is to obtain better models of metabolic pathways. Existing models are dependent on the annotation of the genes. This propagates error accumulation when the pathways are predicted by incorrectly annotated genes. Pairwise classification methods are supervised learning methods used to classify new pair of entities. Some of these classification methods, e.g., Pairwise Support Vector Machines (SVMs), use pairwise kernels. Pairwise kernels describe similarity measures between two pairs of entities. Using pairwise kernels to handle sequence data requires long processing times and large storage. Rational kernels are kernels based on weighted finite-state transducers that represent similarity measures between sequences or automata. They have been effectively used in problems that handle large amount of sequence information such as protein essentiality, natural language processing and machine translations.

Results

We create a new family of pairwise kernels using weighted finite-state transducers (called Pairwise Rational Kernel (PRK)) to predict metabolic pathways from a variety of biological data. PRKs take advantage of the simpler representations and faster algorithms of transducers. Because raw sequence data can be used, the predictor model avoids the errors introduced by incorrect gene annotations. We then developed several experiments with PRKs and Pairwise SVM to validate our methods using the metabolic network of Saccharomyces cerevisiae. As a result, when PRKs are used, our method executes faster in comparison with other pairwise kernels. Also, when we use PRKs combined with other simple kernels that include evolutionary information, the accuracy values have been improved, while maintaining lower construction and execution times.

Conclusions

The power of using kernels is that almost any sort of data can be represented using kernels. Therefore, completely disparate types of data can be combined to add power to kernel-based machine learning methods. When we compared our proposal using PRKs with other similar kernel, the execution times were decreased, with no compromise of accuracy. We also proved that by combining PRKs with other kernels that include evolutionary information, the accuracy can also also be improved. As our proposal can use any type of sequence data, genes do not need to be properly annotated, avoiding accumulation errors because of incorrect previous annotations.     

gamma-Turn types prediction in proteins using the support vector machines

Recently, two different models have been developed for predicting gamma-turns in proteins by Kaur and Raghava [2002. An evaluation of beta-turn prediction methods. Bioinformatics 18, 1508-1514; 2003. A neural-network based method for prediction of gamma-turns in proteins from multiple sequence alignment. Protein Sci. 12, 923-929]. However, the major limitation of previous methods is inability in predicting gamma-turns types. Thus, there is a need to predict gamma-turn types using an approach which will be useful in overall tertiary structure prediction. In this work, support vector machines (SVMs), a powerful model is proposed for predicting gamma-turn types in proteins. The high rates of prediction accuracy showed that the formation of gamma-turn types is evidently correlated with the sequence of tripeptides, and hence can be approximately predicted based on the sequence information of the tripeptides alone.