SA-Search

SA-Search transforms three-dimensional (3D) protein conformations into one-dimensional (1D) Structural Alphabet (SA) sequences using a hidden Markov model and applies sequence-alignment and word-based search methods to identify protein structural similarities.

Key Features:

• HMM-derived Structural Alphabet: Uses a hidden Markov model-derived Structural Alphabet (SA) to encode 3D conformations as a limited set of prototype 1D letters.
• 3D-to-1D compression: Compresses complex 3D structures into simplified 1D representations based on prototype conformations.
• Sequence-alignment-based similarity search: Applies classical sequence alignment methods to SA-encoded sequences to detect 3D structural similarities.
• Exact word extraction (suffix tree): Employs a suffix tree methodology to extract exact structural motifs for precise comparisons.
• Fuzzy word search (1D alignment): Supports identification of fuzzy words by treating them as 1D sequence alignment problems to accommodate conformational variation.
• Rapid 3D similarity searches: Enables rapid and efficient 3D similarity searches by performing analyses on compressed 1D representations.

Scientific Applications:

• Protein folding: Facilitates analysis of protein folding by comparing structural motifs and conformational patterns across proteins.
• Function prediction: Supports function prediction through identification of structural similarity to proteins of known function.
• Evolutionary relationships: Enables inference of evolutionary relationships by detecting conserved structural features across proteins.
• Drug design: Provides structural similarity information relevant to target identification and comparison for drug-design efforts.
• Molecular biology studies: Aids molecular biology research that requires characterization and comparison of protein structural similarity.

Methodology:

Transforms 3D protein conformations into 1D sequences using an HMM-derived Structural Alphabet, compresses structures into prototype conformations, applies classical sequence-alignment methods to SA sequences, uses a suffix tree for exact word extraction, and treats fuzzy word searches as 1D sequence-alignment problems.