sensaas

sensaas aligns and compares molecular shapes and sub-shapes by registering colored 3D point-based representations of van der Waals surfaces to establish correspondences between point clouds.

Key Features:

• Point-Based Representation: Uses 3D point clouds representing the van der Waals surface with each point associated to physico-chemical characteristics.
• Colored Point Labeling: Colors points based on proximity to the nearest atom belonging to user-defined physico-chemical classes to encode chemical information.
• Support for Large Point Clouds: Operates on point clouds containing up to several thousand colored (labeled) points.
• Two-Stage Alignment: Performs an initial global superimposition using pose-invariant local 3D descriptors Fast Point Feature Histograms (FPFH) followed by refinement optimizing correspondence between colored points.
• Rigid-Body Superimposition and Local Shape Focus: Integrates rigid-body superimposition with emphasis on local shape properties to establish accurate correspondences between surfaces.
• Similarity Scoring: Quantifies molecular similarity using Tversky coefficients.

Scientific Applications:

• Benchmarking with X-ray Structures: Reproduces superimposition of X-ray structures from the AstraZeneca (AZ) benchmark data set.
• Comparison with Existing Methods: Enables method comparison and benchmarking against shape-alignment approaches such as shaep and shafts.
• Substructure and Bioisostere Matching: Identifies and aligns substructures and bioisosteric fragments for applications in drug discovery and design.

Methodology:

Registers colored 3D point-based van der Waals surfaces, performs an initial global superimposition using FPFH descriptors, refines matches by optimizing colored-point correspondences via rigid-body superimposition, and evaluates similarity with Tversky coefficients.