Support Vector Machines for Predicting Protein Structural Classes via Pseudo Images Derived From Amino Acid Sequences

Betul Akcesme,Mehmet Can

doi:10.21533/scjournal.v5i1.107

Abstract

Support Vector Machines for Predicting Protein Structural Classes via Pseudo Images Derived From Amino Acid Sequences

Highlights

Support Vector Machine (SVM) is one type of learning machine based on statistical learning theory which has been relatively recently introduced to the field (Vapnik 1995)
SVMs have been applied to solve a variety of problems in the field of bioinformatics such as gene function analysis, microarray expression data (Brown et al 2000), protein secondary structure prediction (Hua and Sun 2001), protein fold recognition (Ding and Dubchak 2001), and cancer tissue classification from microarray expression data (Mukherjee et al 1999)
SVM enables to handle with a large number of features which may causes over-fitting data problem (Vapnik 1998; Cai et al 2003)

Summary

Introduction

Support Vector Machine (SVM) is one type of learning machine based on statistical learning theory which has been relatively recently introduced to the field (Vapnik 1995). Since the first application of SVM for prediction of protein structural classes (Cai et al 2001) it has gained popularity in a wide range of studies. First step is to map the input vectors (protein sequence data) into a feature space with higher dimension, linearly or non-linearly. It seeks an optimized linear division within the feature space from the first step to construct a hyperplane which divides the data points into their corresponding classes. SVM enables to handle with a large number of features which may causes over-fitting data problem (Vapnik 1998; Cai et al 2003)

Methods

Results

Conclusion