Related Experiment Videos
Predicting location and structure of beta-sheet regions using stochastic tree grammars
1Theory NEC Laboratory, RWCP, Kawasaki, Japan.
Proceedings. International Conference on Intelligent Systems for Molecular Biology
|January 1, 1994
Summary
This study introduces a novel method using Stochastic Ranked Node Rewriting Grammars (SRNRG) to predict protein beta-sheet structures. The approach accurately identifies beta-sheet regions in diverse proteins, even with low sequence homology.
Area of Science:
- Computational Biology
- Bioinformatics
- Structural Biology
Background:
- Protein secondary structure prediction is crucial for understanding protein function.
- Identifying beta-sheet regions is particularly challenging due to long-range dependencies.
Purpose of the Study:
- To develop and validate a novel method for predicting protein secondary structures, focusing on beta-sheet regions.
- To leverage Stochastic Ranked Node Rewriting Grammars (SRNRG) for sequence pattern representation.
Main Methods:
- Utilized Stochastic Ranked Node Rewriting Grammars (SRNRG) for amino acid sequence patterns.
- Applied an extended 'Inside-Outside' algorithm with modifications for SRNRGs.
- Reduced amino acid alphabet size by clustering physicochemical properties.
- Parallelized parsing algorithm on a 32-processor CM-5 for efficiency.
Main Results:
- Successfully predicted beta-sheet regions in proteins with <25% sequence homology to training data.
- Demonstrated the effectiveness of SRNRGs in capturing long-distance dependencies.
- Achieved nearly linear speed-up through parallelization.
Conclusions:
- The SRNRG-based method offers an effective and efficient approach for protein secondary structure prediction.
- This method shows promise for predicting beta-sheet structures in proteins with limited sequence similarity.
- The modifications to the learning algorithm and parallelization enhance computational feasibility.