Developing eThread pipeline using SAGA-pilot abstraction for large-scale structural bioinformatics.

Area of Science:

• Structural Bioinformatics
• Computational Biology
• Systems Biology

Background:

• Sequence-based computational annotation is common, but protein threading methods offer valuable structural insights for function prediction.
• Existing threading tools are computationally intensive, limiting their application for large-scale genome analysis.

Purpose of the Study:

• To develop an efficient and scalable pipeline, eThread, for meta-threading protein structure modeling.
• To enable genome-wide structural bioinformatics and structure-based annotation, particularly for prokaryotic genomes.

Main Methods:

• Developed a pilot-based approach supporting data and task-level parallelism for efficient resource utilization.
• Deployed a scalable pipeline on Amazon EC2, enabling dynamic resource selection based on task requirements.
• Conducted runtime analysis to characterize computational complexity and optimize performance.

Main Results:

• The eThread pipeline demonstrates efficient and effective use of computational resources.
• The system scales to support a large number of protein sequences.
• Runtime analysis provides insights for optimizing time-to-solution and cost-to-solution.

Conclusions:

• The eThread pipeline is a viable solution for genome-scale structural bioinformatics and annotation.
• The approach is particularly suitable for analyzing smaller genomes, such as those from prokaryotes.
• The pipeline is extensible to other distributed cyberinfrastructure environments.