Predicting the multi-label protein subcellular localization through multi-information fusion and MLSI dimensionality...

Area of Science:

• Computational biology
• Bioinformatics
• Molecular biology

Background:

• Multi-label (ML) protein subcellular localization (SCL) is crucial for understanding protein function.
• Accurate SCL aids in disease research, including COVID-19.
• Existing methods may face challenges with complex ML SCL prediction.

Purpose of the Study:

• To develop and validate a novel computational method for ML protein SCL prediction.
• To improve the accuracy and efficiency of SCL prediction.
• To provide a tool for analyzing protein localization relevant to diseases like SARS-CoV-2.

Main Methods:

• The ML-locMLFE method integrates six feature extraction techniques.
• Latent semantic indexing is used to reduce redundant information.
• Feature-induced labeling information enrichment is applied for ML prediction.

Main Results:

• ML-locMLFE achieved high accuracy on various datasets (e.g., 99.23% on Gram-positive bacteria).
• The method demonstrated strong performance on the SARS-CoV-2 dataset with 72.73% accuracy.
• Leave-one-out cross-validation confirmed the method's robustness.

Conclusions:

• ML-locMLFE offers significant advantages for ML protein SCL prediction.
• The method provides novel insights for future SCL research.
• This work contributes to the understanding of protein localization in various organisms and viruses.