Sign: large-scale gene network estimation environment for high performance computing

Area of Science:

• Computational Biology
• Bioinformatics
• Systems Biology

Background:

• Gene expression data analysis is crucial for understanding biological systems.
• Estimating large-scale gene networks requires significant computational power.
• Existing methods may not fully leverage high-performance computing resources.

Purpose of the Study:

• To develop and introduce SiGN, a software suite for estimating large-scale gene networks.
• To optimize gene network estimation for high-performance computing environments like the K computer.
• To provide researchers with advanced tools for systems biology analysis.

Main Methods:

• SiGN software suite includes SiGN-BN, SiGN-SSM, and SiGN-L1.
• Employs five distinct models: Bayesian networks (static/dynamic, non-parametric), state space models, graphical Gaussian models, and vector autoregressive models.
• Designed to utilize petaflops-scale computational capabilities for large network inference.

Main Results:

• SiGN is tailored for supercomputers such as the K computer and HGC system.
• The software facilitates the estimation of complex gene networks from gene expression data.
• Provides a scalable solution for computationally intensive network inference tasks.

Conclusions:

• SiGN offers a powerful and efficient approach to large-scale gene network estimation.
• The software enhances systems biology research by enabling analysis on high-performance computing platforms.
• Accessibility for K computer and HGC users promotes broader adoption and research advancement.