SAPH-ire TFx

SAPH-ire TFx predicts the functional significance of experimentally observed post-translational modifications (PTMs) in eukaryotic proteins.

Key Features:

• Artificial Neural Network Model: Employs a multi-feature artificial neural network tailored for functional prediction of experimentally observed eukaryotic PTMs.
• Optimization Metrics: Optimizes model performance using receiver operating characteristic (ROC) and recall metrics.
• Benchmarking and Performance: Benchmarks against independent datasets and alternative models, demonstrating superior recall of known functional PTM sites and recommendation of experimentally confirmed functional PTMs.
• Feature Contribution Analysis: Analyzes feature contributions to demonstrate the necessity of integrating multiple features rather than relying on single-feature approaches using data from public databases.

Scientific Applications:

• Prioritization of PTMs for Experimental Validation: Recommends PTMs with high potential biological impact to aid experimental prioritization, addressing that fewer than ~2% of PTMs have assigned biological functions.
• Insight into Functional Diversity of PTMs: Provides insights into functional diversity to help researchers understand variation in PTM functional equivalence and guide experimental design.

Methodology:

Uses a multi-feature artificial neural network trained and optimized with ROC and recall metrics, benchmarked against independent datasets and alternative models, with analysis of feature contributions.