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

By Chris H. Q. Ding, Inna Dubchak. Bioinform., 2001

893 citations recorded by the source corpus.

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.

Subfields: Molecular Biology

View Paper: https://www.semanticscholar.org/paper/144a0f13ea5fedcc6f142b9845fc99e61fd99eb3

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Molecular Biology

Citations in the corpus

An ensemble approach to protein fold classification by integration of template‐based assignment and support vector machine classifier (2016, methodology): “Third, the majority of previous studies for protein fold classification are based on SCOP (Chen and Kurgan, 2007; Damoulas and Girolami, 2008; Ding and Dubchak, 2001; Dong et al., 2009; Guo and Gao, 2008; Lyons et al., 2015; Rangwala and Karypis, 2005; Shamim et al., 2007; Sharma et al., 2013; Shen ”
Nonlinear Models Using Dirichlet Process Mixtures (2007, methodology): “As a benchmark, we also present the results for the SVM and NN models developed by Ding and Dubchak (2001) on the exact same dataset.”
The Spectrum Kernel: A String Kernel for SVM Protein Classification (2001, background): “Discrimi-native approaches 10 ; 11 ; 12 take a di(cid:11)erent point of view: protein sequences are seen as a set of labeled examples { positive if they are in the family and negative otherwise { and a learning algorithm attempts to learn the distinction between the di(cid:11)erent classes.”
A Unified View on Multi-class Support Vector Classification (2016, methodology): “We considered the LLW and RM machines in our experiments, which are typically omitted in comparison studies for training time reasons (e.g., Ding and Dubchak, 2001; Rifkin and Klautau, 2004; Statnikov et al., 2005), and we could perform proper model selection.”
Prediction of MHC class I binding peptides, using SVMHC (2002, methodology): “Support vector machines are a class of machine learning methods that recently has been applied for classification of microarray data, protein structure prediction and other biological problems [14–16].”
Comparison of machine learning and traditional classifiers in glaucoma diagnosis (2002, methodology): “It has also been applied to data from gene expression [30], DNA and protein analysis [31], [32].”
Probabilistic multi-class multi-kernel learning: on protein fold recognition and remote homology detection (2008, methodology): “In the present work, we concentrate on the same benchmark dataset of Ding and Dubchak (2001) with the extra groups of features proposed by Shen and Chou (2006), and also including sequence-alignment(1) features via a pairwise kernel (Liao and Noble, 2003), which essentially describes the sequence ba”
iFC2: an integrated web-server for improved prediction of protein structural class, fold type, and secondary structure content (2010, methodology): “Numerous computational methods have been proposed for sequence-based prediction of the family and superfamily (Enright et al. 2002; Melvin et al. 2007; Gewehr et al. 2007), the fold type ( Ding and Dubchak 2001; Hvidsten et al. 2003; Shen and Chou 2006; Jeong et al. 2006; Taguchi and Gromiha 2007; ”
Support vector machines for automated gait classification (2005, methodology): “Because of superior gait classification performance as demonstrated in this research and also classification in other biomedical applications [27], support vector machine appears”
REVIEW : Recent advances in developing web-servers for predicting protein attributes (2009, background): “Facing this kind of situation, a different strategy, the so-called taxonomic approach [227] was developed to address the problem.”