Information-Theoretic Evaluation for Computational Biomedical Ontologies

Preface

Contents

1 Introduction

1.1 Background

1.2 Protein Function Prediction Scenarios

1.3 State of the Art Methods

References

2 Methods

2.1 Calculating the Joint Probability of a Graph

2.1.1 Calculating the Information Content of a Graph

2.1.2 Comparing Two Annotation Graphs

2.1.3 Measuring the Quality of Function Prediction

2.1.4 Weighted Metrics

2.1.5 Semantic Distance

2.1.6 Precision and Recall

2.1.7 Supplementary Evaluation Metrics

2.1.8 Additional Topological Metrics

2.2 Confusion Matrix Interpretation of ru and mi

2.3 Annotation Models

2.3.1 The Naïve Model

2.3.2 The BLAST Model

2.3.3 The GOtcha Model

References

3 Experiments and Results

3.1 Average Information Content of a Protein

3.2 Comparative Examples of Calculating Information Content

3.3 Two-Dimensional Plots

3.4 Comparisons of Single Statistics

References

4 Discussion

References

Index