AlphaPred

AlphaPred predicts alpha-turn residues in protein sequences using artificial neural networks trained on sequence-derived evolutionary and secondary structure features.

Key Features:

• Artificial Neural Network Prediction: Uses an artificial neural network (ANN) model trained with position-specific scoring matrices generated by PSI-BLAST and secondary structure predictions from PSIPRED.
• Two-Stage Neural Network Architecture: Implements a feed-forward back-propagation neural network with a single hidden layer consisting of a sequence-structure network followed by a structure-structure network for prediction refinement.
• Sequence-Derived Evolutionary Features: Utilizes PSI-BLAST-generated position-specific scoring matrices derived from multiple sequence alignments to capture evolutionary information.
• Secondary Structure Integration: Incorporates PSIPRED-predicted secondary structure information to improve alpha-turn prediction accuracy.

Scientific Applications:

• Protein Structure Prediction: Identifies alpha-turn residues to support analysis of protein folding, structural motifs, and conformational stability.
• Protein Function Analysis: Facilitates investigation of structural elements influencing protein interactions, enzymatic activity, and molecular recognition.
• Structural Bioinformatics Research: Supports studies of sequence–structure relationships in proteins using computational prediction of local structural motifs.

Methodology:

AlphaPred applies a two-stage feed-forward back-propagation artificial neural network trained on PSI-BLAST-derived position-specific scoring matrices and PSIPRED secondary structure predictions using a dataset of 193 non-homologous protein X-ray structures with five-fold cross-validation.