FUSION

FUSION performs efficient conformational sampling of proteins in continuous space using a fragment-free probabilistic graphical model.

Key Features:

• Fragment-Free Approach: Operates in continuous space without assembling discretized structural fragments from protein structure databases.
• Probabilistic Graphical Model: Employs a probabilistic graphical model to capture complex dependencies and guide conformational sampling.
• Continuous Space Sampling: Samples conformations in continuous space to reduce bottlenecks of fragment-based methods and facilitate convergence for large proteins.
• Performance and Accuracy: In CASP11 blind-target benchmarks, demonstrated superior performance to ROSETTA for proteins exceeding 150 residues and exhibited strong convergence properties.

Scientific Applications:

• Protein Structure Prediction: Predicts tertiary structures of proteins, including large proteins that are challenging for fragment-assembly approaches.
• Protein Folding and Dynamics: Enables exploration of continuous conformational landscapes to support studies of folding pathways and conformational dynamics.
• Function and Interaction Analysis: Provides structural models to investigate protein function and intermolecular interactions.
• Protein Design: Facilitates design of novel proteins by enabling sampling of alternative conformations without fragment constraints.

Methodology:

Uses a probabilistic graphical model to navigate the conformational landscape and performs continuous-space sampling and evaluation of protein conformations without discrete fragment combinations.