Predicting subcellular location of protein with evolution information and sequence-based deep learning.

Area of Science:

• Computational Biology
• Bioinformatics
• Molecular Biology

Background:

• Protein subcellular localization is crucial for understanding biological functions.
• Traditional prediction methods are often laborious and time-consuming.
• Existing machine learning methods frequently overlook protein evolutionary information.

Purpose of the Study:

• To develop an accurate deep learning-based method for protein subcellular localization prediction.
• To integrate evolutionary information into the prediction model.
• To enhance the precision of subcellular location predictions.

Main Methods:

• Utilized a deep learning architecture combining bidirectional long short-term memory (LSTM) and convolutional neural networks (CNN).
• Incorporated amino acid composition, protein sequences, and evolution matrices (Position-Specific Scoring Matrix - PSSM).
• Trained and validated the model on two benchmark datasets.

Main Results:

• Achieved an average precision of 0.7901, ranking loss of 0.0758, and coverage of 1.2848 on benchmark datasets.
• Demonstrated superior performance compared to five existing prediction methods.
• The model effectively extracts features from protein sequences and evolutionary data.

Conclusions:

• The proposed deep learning method offers an accurate and acceptable alternative for protein subcellular localization prediction.
• Integrating evolutionary information significantly improves prediction accuracy.
• The method's performance highlights the importance of sequence and evolutionary data in bioinformatics.