Contents

• Contents
• Introduction
• Methods
  • The problem of objectively testing secondary structure prediction methods
  • Training and test sets of protein structures
  • Generating the multiple sequence alignments
  • Prediction methods analysed
  • Consensus Prediction Method
  • Assessment of accuracy
• Results and Discussion
  • Comparison of secondary structure definition methods
  • Analysis of the test and training alignments
  • Alignment quality
  • Comparison of prediction accuracy for the CB396 test and RS126 training sets
  • Effect on Q₃ of changing the number of related sequences
  • Effect on Q₃ of reducing redundancy in multiple alignments
  • The effect on accuracy of alternative 8 to 3 state reductions
  • Single sequence prediction methods
  • Improving the consensus prediction
• Summary and Conclusions
• Acknowledgements
• Bibliography