RUBI

RUBI predicts ubiquitinated lysine residues in eukaryotic proteomes using sequence-based models to enable genome-scale identification of ubiquitination sites.

Key Features:

• Sequence-based prediction: Uses sequence information and large-scale human ubiquitination datasets to predict lysine ubiquitination across eukaryotic proteomes.
• Iterative model construction: Model development employed iterative optimization focused on factors influencing prediction accuracy.
• Performance metrics: Achieves an Area Under the Curve (AUC) of 0.868 in large cross-validation sets and outperforms existing methods on independent datasets.
• Prediction accuracy: Predicts the number of ubiquitination sites within three sites for approximately 80% of tested proteins.
• Proteome coverage: Estimates that at least 25% of the proteome across multiple model organisms, including thousands of human proteins, may be ubiquitinated.
• Cross-species applicability: Predictive signal is consistent across different eukaryotic organisms, supporting comparative and evolutionary analyses.
• Influence of intrinsic disorder: Reports a weak anti-correlation between predicted intrinsic disorder and ubiquitination, with residues of low disorder probability being preferred for ubiquitination.

Scientific Applications:

• Functional genomics: Identifying potential regulatory ubiquitination sites that may modulate protein function and cellular processes.
• Proteomics: Predicting post-translational modification sites to inform proteome dynamics and experimental prioritization.
• Drug discovery: Prioritizing specific ubiquitination sites as potential targets for therapeutic intervention in diseases involving dysregulated protein ubiquitination.

Methodology:

Sequence-based model development using large-scale human ubiquitination datasets; iterative model construction and optimization; evaluation with large cross-validation sets and independent dataset testing; incorporation of predicted intrinsic protein disorder as a feature.