Bingham deep neural and oppositional fish swarm optimized protein structure prediction.

Area of Science:

• Biochemistry
• Bioinformatics
• Computational Biology

Background:

• Essential proteins manage cellular activities, and their structure prediction aids in understanding cellular functions.
• Existing protein prediction methods struggle with imbalanced data, leading to unsatisfactory sensitivity.
• Accurate prediction of protein structure and identification of essential proteins are crucial for biological research.

Purpose of the Study:

• To develop a novel method for precise secondary protein structure prediction and essential protein identification.
• To address the challenge of imbalanced characteristics in existing prediction methods.
• To improve the accuracy and sensitivity of protein structure prediction.

Main Methods:

• A Bingham Distributed Deep Convolutional (BDDC) framework was designed to identify essential proteins, mitigating imbalanced learning issues.
• An Oppositional Artificial Fish Swarm Optimization framework was proposed for precise secondary structure prediction.
• The method emulates artificial fish behaviors (foraging, following, swarming) using proximal count, oppositional, and Gaussian functions.

Main Results:

• The BDC-OAFO method demonstrated superior performance in identifying essential proteins compared to existing methods.
• Experimental results on the Protein Data Bank dataset confirmed the method's effectiveness in precise secondary protein structure prediction.
• The proposed approach significantly improved prediction accuracy, particularly in handling imbalanced datasets.

Conclusions:

• The BDC-OAFO method offers a significant advancement in essential protein identification and secondary structure prediction.
• The novel approach effectively overcomes the limitations of low sensitivity caused by imbalanced data characteristics.
• BDC-OAFO provides a more reliable tool for comprehending cellular functions through accurate protein analysis.