Scalable Pattern Recognition Algorithms : Applications in Computational Biology and Bioinformatics

This book addresses the need for a unified framework describing how soft computing and machine learning techniques can be judiciously formulated and used in building efficient pattern recognition models. The text reviews both established and cutting-edge research, providing a careful balance of theory, algorithms, and applications, with a particular emphasis given to applications in computational biology and bioinformatics. Features: integrates different soft computing and machine learning methodologies with pattern recognition tasks; discusses in detail the integration of different techniques for handling uncertainties in decision-making and efficiently mining large biological datasets; presents a particular emphasis on real-life applications, such as microarray expression datasets and magnetic resonance images; includes numerous examples and experimental results to support the theoretical concepts described; concludes each chapter with directions for future research and a comprehensive bibliography. Foreword......Page 6 Preface......Page 9 Contents......Page 15 1.1 Introduction......Page 21 1.2.1 Nucleic Acids......Page 23 1.2.2 Proteins......Page 25 1.3.1 Alignment and Comparison of DNA, RNA, and Protein Sequences......Page 26 1.3.2 Identification of Genes and Functional Sites from DNA Sequences......Page 27 1.3.3 Prediction of Protein Functional Sites......Page 28 1.3.5 Protein Structure Prediction and Classification......Page 29 1.3.6 Molecular Design and Molecular Docking......Page 30 1.3.8 Analysis of Microarray Expression Data......Page 31 1.4 Pattern Recognition Perspective......Page 35 1.4.1 Pattern Recognition......Page 36 1.4.2 Relevance of Soft Computing......Page 40 1.5 Scope and Organization of the Book......Page 42 References......Page 46 Part I Classification......Page 63 2.1 Introduction......Page 64 2.2 Neural Network Based Tree-Structured Pattern Classifier......Page 66 2.2.1 Selection of Multilayer Perceptron......Page 68 2.2.2 Splitting and Stopping Criteria......Page 69 2.3 Identification of Splice-Junction in DNA Sequence......Page 70 2.3.2 Experimental Results......Page 71 2.4 Identification of Protein Coding Region in DNA Sequence......Page 72 2.4.1 Data and Method......Page 75 2.4.2 Feature Set......Page 76 2.4.3 Experimental Results......Page 78 References......Page 83 3.1 Introduction......Page 86 3.2.1 Bio-Basis Function......Page 88 3.2.2 Selection of Bio-Basis Strings Using Mutual Information......Page 91 3.2.3 Selection of Bio-Basis Strings Using Fisher Ratio......Page 93 3.3.1 Asymmetricity of Biological Dissimilarity......Page 94 3.3.2 Novel Bio-Basis Function......Page 95 3.4 Biological Dissimilarity Based String Selection Method......Page 96 3.4.1 Fisher Ratio Using Biological Dissimilarity......Page 97 3.4.2 Nearest Mean Classifier......Page 99 3.4.3 Degree of Resemblance......Page 100 3.4.4 Details of the Algorithm......Page 101 3.5.1 Compactness: α Index......Page 102 3.5.3 Class Separability: γ Index......Page 103 3.6 Experimental Results......Page 104 3.6.1 Support Vector Machine......Page 105 3.6.2 Description of Data Set......Page 106 3.6.3 Illustrative Example......Page 108 3.6.4 Performance of Different String Selection Methods......Page 109 3.6.5 Performance of Novel Bio-Basis Function......Page 117 3.7 Conclusion and Discussion......Page 118 References......Page 119 Part II Feature Selection......Page 121 4.1 Introduction......Page 122 4.2 Basics of Rough Sets......Page 125 4.3 Rough Set-Based Molecular Descriptor Selection Algorithm......Page 128 4.3.1 Maximum Relevance-Maximum Significance Criterion......Page 129 4.3.3 Generation of Equivalence Classes......Page 131 4.4.1 Description of QSAR Data Sets......Page 132 4.4.2 Support Vector Regression Method......Page 133 4.4.4 Performance Analysis......Page 134 4.4.5 Comparative Performance Analysis......Page 139 4.5 Conclusion and Discussion......Page 142 References......Page 143 5.1 Introduction......Page 147 5.2 Gene Selection Using f-Information Measures......Page 149 5.2.1 Minimum Redundancy-Maximum Relevance Criterion......Page 150 5.2.2 f-Information Measures for Gene Selection......Page 151 5.3 Experimental Results......Page 154 5.3.2 Class Prediction Methods......Page 155 5.3.3 Performance Analysis......Page 156 5.3.4 Analysis Using Class Separability Index......Page 160 5.4 Conclusion and Discussion......Page 165 References......Page 166 6.1 Introduction......Page 170 6.2 Integrated Method for Identifying Disease Genes......Page 172 6.3 Experimental Results......Page 174 6.3.3 Overlap with Known Disease-Related Genes......Page 175 6.3.4 PPI Data and Shortest Path Analysis......Page 178 6.3.5 Comparative Performance Analysis of Different Methods......Page 180 References......Page 182 7.1 Introduction......Page 186 7.2 Selection of Differentially Expressed miRNAs......Page 189 7.2.1 RSMRMS Algorithm......Page 190 7.2.2 Fuzzy Discretization......Page 191 7.2.3 B.632+ Error Rate......Page 194 7.3.1 Data Sets Used......Page 195 7.3.2 Optimum Values of Different Parameters......Page 196 7.3.3 Importance of B.632+ Error Rate......Page 197 7.3.4 Role of Fuzzy Discretization Method......Page 200 7.3.5 Comparative Performance Analysis......Page 201 7.4 Conclusion and Discussion......Page 204 References......Page 206 Part III Clustering......Page 209 8.1 Introduction......Page 210 8.2.1 Different Gene Clustering Algorithms......Page 213 8.2.2 Quantitative Measures......Page 218 8.3.1 Rough--Fuzzy C-Means......Page 220 8.3.2 Initialization Method......Page 223 8.3.3 Identification of Optimum Parameters......Page 224 8.4.1 Gene Expression Data Sets Used......Page 225 8.4.2 Optimum Values of Different Parameters......Page 226 8.4.3 Importance of Correlation-Based Initialization Method......Page 227 8.4.6 Eisen Plots......Page 229 8.4.7 Biological Significance Analysis......Page 230 8.4.8 Functional Consistency of Clustering Result......Page 233 References......Page 234 9.1 Introduction......Page 238 9.2.1 Gene Clustering: Supervised Versus Unsupervised......Page 240 9.2.2 Criteria for Gene Selection and Clustering......Page 241 9.3.1 Supervised Similarity Measure......Page 242 9.3.2 Gene Clustering Algorithm......Page 245 9.3.4 Computational Complexity......Page 248 9.4.1 Gene Expression Data Sets Used......Page 249 9.4.2 Optimum Value of Threshold......Page 250 9.4.3 Qualitative Analysis of Supervised Clusters......Page 251 9.4.4 Importance of Supervised Similarity Measure......Page 252 9.4.5 Importance of Augmented Genes......Page 253 9.4.6 Performance of Coarse and Finer Clusters......Page 256 9.4.7 Comparative Performance Analysis......Page 259 9.5 Conclusion and Discussion......Page 262 References......Page 263 10.1 Introduction......Page 266 10.2.1 Basics of Biclustering......Page 269 10.2.2 Possibilistic Clustering......Page 271 10.3.1 Objective Function......Page 272 10.3.2 Bicluster Means......Page 274 10.3.3 Convergence Condition......Page 275 10.3.4 Details of the Algorithm......Page 276 10.3.6 Selection of Initial Biclusters......Page 278 10.4.1 Average Number of Genes......Page 279 10.4.4 Average Mean Squared Residue......Page 280 10.5 Experimental Results......Page 281 10.5.1 Optimum Values of Different Parameters......Page 282 10.5.2 Analysis of Generated Biclusters......Page 283 10.5.3 Comparative Analysis of Different Methods......Page 285 10.6 Conclusion and Discussion......Page 286 References......Page 287 11.1 Introduction......Page 290 11.2.1 Fuzzy Set......Page 292 11.2.2 Co-Occurrence Matrix......Page 293 11.2.4 Second Order Fuzzy Entropy......Page 294 11.2.6 2D S-Type Membership Function......Page 295 11.3.1 Modification of Co-Occurrence Matrix......Page 296 11.3.2 Measure of Ambiguity......Page 298 11.3.3 Strength of Ambiguity......Page 299 11.4 Experimental Results......Page 304 References......Page 308 About the Authors......Page 311 Index......Page 313 Recent advances in high-throughput technologies have resulted in a deluge of biological information. Yet the storage, analysis, and interpretation of such multifaceted data require effective and efficient computational tools. This unique text/reference addresses the need for a unified framework describing how soft computing and machine learning techniques can be judiciously formulated and used in building efficient pattern recognition models. The book reviews both established and cutting-edge research, following a clear structure reflecting the major phases of a pattern recognition system: classification, feature selection, and clustering. The text provides a careful balance of theory, algorithms, and applications, with a particular emphasis given to applications in computational biology and bioinformatics. Topics and features: Reviews the development of scalable pattern recognition algorithms for computational biology and bioinformatics Integrates different soft computing and machine learning methodologies with pattern recognition tasks Discusses in detail the integration of different techniques for handling uncertainties in decision-making and efficiently mining large biological datasets Presents a particular emphasis on real-life applications, such as microarray expression datasets and magnetic resonance images Includes numerous examples and experimental results to support the theoretical concepts described Concludes each chapter with directions for future research and a comprehensive bibliography This important work will be of great use to graduate students and researchers in the fields of computer science, electrical and biomedical engineering. Researchers and practitioners involved in pattern recognition, machine learning, computational biology and bioinformatics, data mining, and soft computing will also find the book invaluable