ClubSub-P: Cluster-Based Subcellular Localization Prediction for Gram-Negative Bacteria and Archaea

Area of Science:

• Computational biology
• Bioinformatics
• Microbial genomics

Background:

• Protein subcellular localization (SCL) is crucial for understanding cellular function.
• Existing SCL prediction tools suffer from high false-positive/negative rates and misannotated start codons.
• Accurate SCL prediction is vital for identifying potential vaccine targets, particularly in Gram-negative bacteria.

Purpose of the Study:

• To develop an improved method for predicting protein subcellular localization.
• To address limitations of current SCL prediction tools, including annotation errors.
• To create a searchable online database for SCL prediction in microorganisms.

Main Methods:

• Developed ClubSub-P, an online database integrating multiple SCL prediction tools into a consensus pipeline.
• Utilized over 600 fully sequenced microbial proteomes.
• Applied clustering of homologous proteins from Gram-negative bacteria and Archaea to refine predictions.

Main Results:

• ClubSub-P achieves high precision in assigning SCL for proteins from Gram-negative bacteria and Archaea.
• The consensus approach effectively eliminates false-positive and false-negative predictions.
• Identified and corrected misannotated start codons and other annotation errors.

Conclusions:

• ClubSub-P offers a robust and accurate solution for protein SCL prediction.
• The database facilitates improved identification of vaccine targets and enhances genomic annotation quality.
• The platform is expandable with new genomic data and prediction tools, promising continued performance improvements.