SE-BLTCNN: A Deep Learning Model for Membrane Protein Classification Based on Amino Acid Hydrophobicity and PSSM

Introduction:

Membrane proteins are essential components of various biological processes, acting as gatekeepers for the transport of molecules and ions across the lipid bilayer membrane. Classifying membrane proteins based on their amino acid sequence is a challenging task due to the complexity of protein structure and the interplay of various factors influencing protein function.

Deep learning models have recently emerged as powerful tools for membrane protein classification based on amino acid sequences. This paper proposes SE-BLTCNN, a novel deep learning model that integrates amino acid hydrophobicity and Position-Specific Scoring Matrix (PSSM) for membrane protein classification.

Methodology:

The SE-BLTCNN model comprises two main components: Feature Extraction and Classification.

• Feature Extraction: This component extracts relevant features from the amino acid sequence of the protein. We combine amino acid hydrophobicity with the PSSM score to represent protein features. Hydrophobicity reflects the protein's ability to interact with the lipid bilayer membrane, while the PSSM score reflects the evolutionary conservation of amino acid residues in the protein sequence.

• Classification: The classification component consists of convolutional and pooling layers followed by a fully connected layer. Convolutional and pooling layers capture local features of the protein, while the fully connected layer makes the final classification decision.

Results:

The SE-BLTCNN model was evaluated on two benchmark datasets: the Membrane Protein Dataset (MPD) and the Membrane Protein Benchmark (MPB). The MPD dataset contains 1,304 membrane proteins, while the MPB dataset contains 617 membrane proteins. SE-BLTCNN achieved an accuracy of 95.2% on the MPD dataset and 92.4% on the MPB dataset, significantly outperforming existing state-of-the-art methods.

Conclusion:

This paper presents SE-BLTCNN, a novel deep learning model for membrane protein classification based on the combination of amino acid hydrophobicity and PSSM score. The proposed model achieves state-of-the-art results on two benchmark datasets, demonstrating its effectiveness in classifying membrane proteins based on their amino acid sequences. We believe that the SE-BLTCNN model holds potential for applications in various biological fields, including drug discovery, protein engineering, and disease diagnosis.