BitQT

BitQT clusters molecular dynamics (MD) trajectories using a graph-based Quality Threshold (QT) approach to identify conformational ensembles that satisfy a user-defined collective similarity threshold.

Key Features:

• Graph transformation: Transforms MD trajectories into graphs where nodes represent conformations and unweighted edges indicate mutual similarity.
• Binary-encoded RMSD matrix: Constructs a binary-encoded Root Mean Square Deviation (RMSD) matrix to represent pairwise similarity.
• Bitwise operations: Performs cluster extraction using bitwise operations on the binary RMSD matrix to improve computational efficiency.
• Maximum Clique mapping: Maps QT clustering to the Maximum Clique Problem from graph theory to enable efficient cluster identification.
• Quality Threshold enforcement: Enforces the QT constraint so all cluster members meet the user-defined collective similarity.
• Consistency with exact QT: Produces results consistent with exact QT clustering methods while reducing computational cost.
• Scalability: Reduces computational demands to scale to large and sparse MD datasets.

Scientific Applications:

• Conformational ensemble identification: Identifying conformational ensembles from MD trajectories by grouping conformations that meet similarity thresholds.
• Downstream analysis reduction: Simplifying downstream analyses of MD simulations by reducing large trajectory datasets into representative clusters.
• Method comparison: Providing a computationally efficient alternative for comparison against exact QT clustering results.

Methodology:

MD trajectories are transformed into graphs (nodes = conformations, unweighted edges = mutual similarity) via a binary-encoded RMSD matrix; clusters are extracted using bitwise operations and by framing QT clustering as a Maximum Clique Problem while enforcing a user-defined collective similarity threshold.