APPAGATO: an APproximate PArallel and stochastic GrAph querying TOol for biological networks

Area of Science:

• Bioinformatics
• Computational Biology
• Systems Biology

Background:

• Biological network analysis is computationally intensive.
• Existing tools struggle with large, noisy biological network data.
• Approximate network matching is crucial for handling data imperfections.

Purpose of the Study:

• To present APPAGATO, a novel stochastic and parallel algorithm for biological network querying.
• To address the computational challenges of finding approximate network matches in large datasets.
• To improve the accuracy and performance of biological network analysis tools.

Main Methods:

• Developed a stochastic and parallel algorithm named APPAGATO.
• Implemented the algorithm using the CUDA-C++ Toolkit 7.0 framework.
• Handled node, edge, and node label mismatches for approximate matching.

Main Results:

• APPAGATO demonstrates higher performance compared to existing tools.
• The algorithm provides statistically significant more accurate results.
• Successfully applied to large protein-protein interaction networks with synthetic and real annotations.

Conclusions:

• APPAGATO is an efficient and accurate solution for querying large biological networks.
• The stochastic and parallel nature of APPAGATO enables scalability.
• The tool facilitates meaningful analysis of noisy biological data and supports cross-species/tissue comparisons.