Determining nucleolar association from sequence by leveraging protein-protein interactions.

Area of Science:

• Cell Biology
• Bioinformatics
• Systems Biology

Background:

• Intra-nuclear organization of proteins is essential for cellular function.
• Modeling intranuclear protein transport is challenging due to passive diffusion and complex molecular interactions.
• Growing data on protein destinations and interactions necessitates integrative modeling approaches.

Purpose of the Study:

• To develop an accurate data-driven model for intranuclear protein trafficking.
• To leverage existing evidence, including genomic sequence and protein-protein interaction data, for improved predictive modeling.
• To specifically enhance the prediction of nucleolar-associated proteins.

Main Methods:

• Utilized Kernel Canonical Correlation Analysis (KCCA) for predictor construction.
• Integrated genomic sequence data with other knowledge sources during KCCA training.
• Focused on inducing protein sequence features and relations relevant to nucleolar protein-protein interactions.

Main Results:

• Achieved approximately 78% success rate in classifying nucleolar association.
• KCCA-induced features outperformed baseline approaches in predicting nucleolar association.
• Coalescing protein-protein interaction data with sequence data improved prediction of key ribosomal and RNA-related nucleolar proteins.

Conclusions:

• The integrative method using KCCA effectively models intranuclear trafficking.
• Combining diverse data sources significantly enhances the prediction of protein associations within the nucleolus.
• This approach aids in understanding the organization of critical nucleolar protein networks.