Wang-xiaoheng

Wang-xiaoheng predicts protein structural classes using fused pseudo amino acid compositions and two-dimensional (2-D) wavelet denoising to improve feature representation for structure-function analysis and rational drug design.

Key Features:

• Wavelet Denoising Methodology: Employs a two-dimensional (2-D) wavelet denoising algorithm to reduce redundancy in feature vectors derived from proteins.
• Innovative Fusion Strategy: Introduces a "first fuse, then denoise" strategy by initially combining two types of pseudo amino acid compositions into feature vectors and subsequently applying 2-D wavelet denoising.
• Pseudo Amino Acid Compositions (PseAAC): Extracts sequence-derived features using two distinct PseAAC methods.
• PWD-FU-PseAAC Model: Integrates parallel 2-D wavelet-denoised feature vectors into a cohesive predictive framework for structural class assignment.
• Experimental Validation: Validated on three low-similarity datasets with reported superior performance in protein structural class prediction compared to traditional methods.

Scientific Applications:

• Protein Functional Analysis: Infers potential protein functions by predicting structural classes that correlate with functional annotations.
• Protein Folding Recognition: Aids in elucidating protein folding patterns and structural stability through class-level assignment.
• Rational Drug Design: Supports identification of structural class–related target sites to inform therapeutic development.

Methodology:

Feature extraction uses two types of PseAAC to generate initial vectors; a 2-D wavelet denoising algorithm is applied to feature vectors; denoised feature vectors are fused into a single comprehensive vector using the "first fuse, then denoise" approach; the PWD-FU-PseAAC model is applied to predict protein structural classes.