ClathPLM: Deep multi-view feature extraction with CNN and attention enhances clathrin protein identification

Area of Science:

• Biochemistry
• Computational Biology
• Molecular Biology

Background:

• Clathrin is crucial for intracellular vesicle transport, particularly clathrin-mediated endocytosis (CME).
• Dysfunctional Clathrin is implicated in diseases like neurodegeneration, cancer metastasis, and immune disorders.
• Existing experimental methods for Clathrin detection are costly and time-consuming.

Purpose of the Study:

• To develop an efficient and reliable computational method for Clathrin recognition.
• To assist in the identification of Clathrin's role in cellular regulation and disease.

Main Methods:

• Proposed ClathPLM, a model integrating sequence embeddings from ProtT5, ProtBert, and ESM-3 pre-trained protein language models.
• Employed deep representation learning using Convolutional Neural Networks (CNN) and Multi-Head Attention (MHA) mechanisms.
• Fused multi-view representations for final classification and evaluated fusion strategies and attention mechanisms.

Main Results:

• ClathPLM achieved excellent classification performance and robustness, outperforming existing state-of-the-art methods.
• The model demonstrated strong performance on a case-study dataset and scalability on a vesicular transport proteins (VTPs) dataset.

Conclusions:

• ClathPLM offers an effective computational approach for Clathrin recognition.
• The model has the potential to advance understanding of Clathrin's role in cellular processes and disease.
• Facilitates future biological research and potential clinical applications.