nhKcr

nhKcr predicts lysine crotonylation (Kcr) sites on nonhistone proteins to enable computational identification of post-translational crotonylation events and to complement high-resolution mass spectrometry-based discovery.

Key Features:

• Specialization: Predicts lysine crotonylation (Kcr) sites exclusively on nonhistone proteins.
• Deep learning framework (CNNrgb): Implements the CNNrgb deep learning-based computational framework for site prediction.
• Feature integration: Integrates diverse types of features for enhanced predictive accuracy.
• Benchmarking: Performance was compared against random forest, logitboost, naïve Bayes, and logistic regression classifiers.
• Validation: Evaluated using 10-fold cross-validation and independent testing.
• Scalability: Demonstrated high computational efficiency on large datasets.
• Method comparison foundation: Development was informed by a review of six existing crotonylation site prediction methods.

Scientific Applications:

• Annotation of Kcr sites: Annotates lysine crotonylation sites on nonhistone proteins for downstream biological and pathophysiological studies.
• Experimental prioritization: Prioritizes candidate Kcr sites for validation by high-resolution mass spectrometry.
• Large-scale sequence analysis: Enables high-throughput analysis of large nonhistone protein sequence datasets to identify potential crotonylation sites.

Methodology:

Implements the CNNrgb deep learning framework integrating diverse feature types; development was informed by a review of six existing predictors and performance was benchmarked against random forest, logitboost, naïve Bayes, and logistic regression using 10-fold cross-validation and independent testing.