DeepTESR

DeepTESR predicts translational elongation short ramp (TESR) scores from mRNA sequences using deep learning trained on ribosome profiling data to quantify translational control and efficiency.

Key Features:

• Deep Learning Architecture: Employs a deep learning model trained on mRNA sequence features that achieved a mean absolute error (MAE) of 0.285 and a coefficient of determination (R²) of 0.627.
• Comparative Performance: Outperforms traditional machine learning models such as LightGBM, which reported MAE 0.335 and R² 0.571 in comparative evaluations.
• Experimental Validation: Predictions were validated by heterologous fluorescence expression assays of proteins with randomly selected TESRs, showing a moderate correlation between predicted and observed translational efficiencies.
• Genome-Wide Analysis Capability: Applied to 4,305 coding sequences from Escherichia coli, revealing conserved TESRs across clusters of orthologous groups.

Scientific Applications:

• Gene Expression Control: Predicts TESR scores to aid interpretation of translational-level regulation of gene expression.
• Translational Mechanism Deciphering: Provides insights into how translational elongation rates influence protein synthesis efficiency.
• Biotechnological Applications: Informs optimization of gene expression systems in synthetic biology and biotechnology by predicting TESR-related effects on protein production.

Methodology:

Utilizes ribosome profiling data and trains a deep learning model on 226,981 TESR sequences to correlate mRNA sequence features with translational elongation rates.