Efficient dynamic programming algorithm with prior knowledge for protein β-strand alignment.

Area of Science:

• Computational Biology
• Structural Bioinformatics
• Protein Structure Prediction

Background:

• Accurate prediction of protein beta-sheet structure relies on effective alignment of beta-strands.
• Previous methods using contact maps for alignment scores suffer from low precision and disregard specific beta-residue structural features.

Purpose of the Study:

• To develop an improved algorithm for beta-strand alignment that incorporates beta-structure information.
• To enhance the accuracy and efficiency of protein beta-sheet structure prediction.

Main Methods:

• Utilized beta-structure information derived from protein datasets as alignment scores.
• Integrated predicted contact maps as prior knowledge to refine alignment scores.
• Modified dynamic programming and applied the Four Russians method to optimize the alignment algorithm.

Main Results:

• The proposed method demonstrated significant improvements in execution time and accuracy of beta-strand alignment.
• Applied to state-of-the-art methods, it led to enhanced beta-sheet structure prediction accuracy.
• Experimental results on the BetaSheet916 dataset validated the method's effectiveness.

Conclusions:

• The novel approach effectively combines beta-structure information and contact map priors for superior beta-strand alignment.
• This advancement contributes to more accurate and computationally efficient protein beta-sheet structure prediction.