Related Experiment Videos
Analysis and prediction of functional sub-types from protein sequence alignments.
1Bioinformatics Research Group, SmithKline Beecham Pharmaceuticals Research & Development, 709 Swedeland Road, King of Prussia, PA 19406, USA.
Journal of Molecular Biology
|October 7, 2000
Summary
This study introduces a novel method for predicting protein functional sub-types using multiple sequence alignments. The approach accurately identifies key positions and classifies uncharacterized proteins, outperforming existing methods.
Area of Science:
- Computational Biology
- Bioinformatics
- Protein Science
Background:
- The expanding scale of protein sequence data necessitates advanced methods for functional characterization.
- Existing methods for predicting protein functional sub-types often struggle with accuracy, especially for distantly related sequences.
Purpose of the Study:
- To develop and validate a novel computational method for analyzing multiple protein sequence alignments to predict functional sub-types.
- To identify sequence positions indicative of functional divergence within protein families.
- To assess the method's predictive accuracy compared to established techniques like sequence similarity and BLAST.
Main Methods:
- Analysis of sub-type-specific sequence profiles and positional entropy in multiple sequence alignments.
- Identification of alignment positions with high relative entropy correlating with known functional sub-type determinants.
- Development and assessment of prediction algorithms using cross-validation, excluding close homologues to simulate real-world scenarios.
Main Results:
- The developed method accurately identifies positions associated with functional sub-types across diverse protein families (e.g., kinases, proteases).
- The prediction method achieves high accuracy (96%) for classifying sequences with known sub-types, significantly outperforming sequence similarity (80%) and BLAST (74%) at 30% sequence identity.
- Large-scale assessment using PFAM and SWISSPROT data confirms superior performance, with an average accuracy of 94% compared to 68% for sequence similarity and 79% for BLAST.
Conclusions:
- The proposed method offers a robust and accurate approach for predicting protein functional sub-types from sequence alignments.
- It provides valuable insights into sequence positions driving functional specificity, guiding experimental design.
- The method has significant implications for improving genome annotation and predicting protein function, even for distantly related sequences.