Identification of self-interacting proteins by exploring evolutionary information embedded in PSI-BLAST-constructed...

Area of Science:

• Computational biology
• Proteomics
• Bioinformatics

Background:

• Self-interacting proteins (SIPs) are crucial for numerous biological processes, including gene regulation and immune response.
• Experimental identification of SIPs is limited, necessitating computational approaches.
• Predicting SIPs from protein sequences is vital for advancing biological research.

Purpose of the Study:

• To develop an effective computational method for predicting SIPs based on protein sequences.
• To introduce the RVMBIGP model, combining Relevance Vector Machine (RVM) and Bi-gram probability (BIGP).

Main Methods:

• Feature extraction using Bi-gram probability (BIGP) on Position Specific Scoring Matrix (PSSM).
• Dimensionality reduction and noise reduction using Principal Component Analysis (PCA).
• Classification using the Relevance Vector Machine (RVM) model.

Main Results:

• The RVMBIGP model achieved high prediction accuracies: 95.48% for yeast and 98.80% for human datasets.
• The method effectively integrates sequence information and robust classification.
• Experimental results demonstrate the model's high performance.

Conclusions:

• The RVMBIGP approach is a promising and cost-effective computational tool for identifying SIPs.
• This method can facilitate future proteomics research.
• The RVMBIGP server is available for academic use.