Rapid protein fragment search using hash functions based on the Fourier transform

Area of Science:

• Structural bioinformatics
• Computational biology
• Biophysics

Background:

• The exponential growth of protein structure databases necessitates efficient similarity search methods.
• Identifying similar protein structures is crucial for understanding protein function and evolution.

Purpose of the Study:

• To develop a novel and efficient method for searching similar three-dimensional protein structure fragments.
• To theoretically prove the method's effectiveness in preserving structural similarity through hash vector comparisons.

Main Methods:

• Associating a hash vector with each fixed-length protein fragment based on C alpha atom distances.
• Utilizing low-frequency components of a Fourier-like spectrum for hash vector generation.
• Theoretically proving that small root mean square distances between fragments correlate with small distances between their hash vectors.

Main Results:

• The proposed hash vector method enables rapid analysis of protein fragment similarity.
• Variants of the method demonstrated significant speed improvements: 18-80 times faster than naive methods and 3-10 times faster than previous approaches.
• Empirical validation using Protein Data Bank (PDB) data confirmed the method's efficiency and accuracy.

Conclusions:

• The novel hash vector method offers a computationally efficient solution for protein structure fragment searching.
• This technique is well-suited for analyzing large and rapidly expanding structural biology datasets.
• The theoretical guarantees ensure reliable similarity assessments, advancing structural bioinformatics research.