Grain protein function prediction based on self-attention mechanism and bidirectional LSTM

Area of Science:

• Bioinformatics and Computational Biology
• Genomics and Proteomics
• Agricultural Science

Background:

• Genome sequencing advancements necessitate efficient bioinformatics tools for predicting protein function.
• Understanding grain protein function is crucial for crop improvement and food security.
• Existing computational methods may not fully leverage diverse protein characteristics.

Purpose of the Study:

• To develop a novel and accurate computational method for predicting grain protein function.
• To improve upon existing neural network algorithms for protein function prediction.
• To validate the proposed method using protein data from key agricultural grains.

Main Methods:

• Proposed a novel neural network algorithm named Chemical-SA-BiLSTM.
• Integrated protein chemical properties with amino acid sequences.
• Combined self-attention mechanism with bidirectional Long Short-Term Memory (BiLSTM) networks.
• Utilized protein data from soybean, maize, indica, and japonica.

Main Results:

• The Chemical-SA-BiLSTM algorithm demonstrated superior performance compared to classical neural network algorithms.
• Achieved higher accuracy in predicting grain protein function.
• Effectiveness validated across diverse grain protein datasets.

Conclusions:

• The Chemical-SA-BiLSTM algorithm is a highly effective tool for grain protein function prediction.
• Integrating chemical properties enhances the accuracy of protein function prediction models.
• This method offers a significant advancement in bioinformatics for agricultural applications.