Robust and accurate prediction of self-interacting proteins from protein sequence information by exploiting weighted...

Area of Science:

• Bioinformatics
• Computational Biology
• Protein Science

Background:

• Self-interacting proteins (SIPs) are crucial for cellular functions.
• Experimental identification of SIPs is time-consuming, costly, and prone to errors.
• Developing automated prediction methods is essential to supplement experimental approaches.

Purpose of the Study:

• To develop an efficient and accurate computational framework for predicting SIPs from protein sequences.
• To leverage protein evolutionary information for improved SIP prediction accuracy.

Main Methods:

• Protein sequences were converted into Position Specific Scoring Matrices (PSSM) using PSI-BLAST.
• Gray Level Co-occurrence Matrix (GLCM) was used for feature extraction from PSSM.
• Adaptive Synthetic (ADASYN) technique balanced the dataset for improved classification.
• Weighted Sparse Representation based Classification (WSRC) model was employed for prediction.

Main Results:

• The GLCM-WSRC framework demonstrated high prediction performance.
• Achieved 98.10% accuracy on the yeast dataset.
• Achieved 91.51% accuracy on the human dataset.

Conclusions:

• The proposed GLCM-WSRC model is an effective tool for automated SIP prediction.
• The method shows potential for large-scale SIP identification and other bioinformatics tasks.
• This approach offers a valuable supplement to experimental methods in protein interaction studies.