ISAC

ISAC accelerates 2D classification of single-particle electron microscopy images to produce high-quality class averages for structural biology.

Key Features:

• GPU acceleration: Implements GPU processing using cards such as Nvidia GeForce GTX 1080 TI to dramatically reduce runtime, achieving performance comparable to twelve high-end cluster nodes with two consumer-grade GPUs and enabling processing of ~1,000,000 particles in approximately 6–13 hours depending on dataset quality and box size.
• Iterative Stable Alignment and Clustering: Executes the ISAC (Iterative Stable Alignment and Clustering) algorithm to perform iterative alignment and clustering of 2D particle images.
• Scalability: Scales linearly across input dimensions to accommodate increasing data volumes from large-scale EM datasets.
• High-quality class averages: Produces robust, high-quality 2D class averages from large single-particle datasets while maintaining accuracy.

Scientific Applications:

• Structural biology: Enables generation of high-quality 2D class averages to support visualization and analysis of macromolecular structures from single-particle electron microscopy datasets.

Methodology:

Implements the Iterative Stable Alignment and Clustering (ISAC) algorithm on GPUs to perform iterative alignment and clustering of 2D single-particle EM images.