Analysis of computational codon usage models and their association with translationally slow codons

Area of Science:

• Molecular Biology
• Computational Biology
• Genomics

Background:

• Accurate prediction of protein translation rates is crucial for understanding co-translational folding.
• Synonymous codons translate at different rates, necessitating computational models that analyze codon usage bias (CUB).
• Existing models use genome-wide CUB or CUB from highly expressed genes to infer translation rates.

Purpose of the Study:

• To evaluate five computational models for their ability to predict translational slowdowns.
• To compare model predictions against ribosome footprint (RFP) data from S. cerevisiae.
• To assess the reliability of RFP data for estimating translation rates.

Main Methods:

• Evaluated five distinct computational translation models.
• Utilized two independent ribosome footprint (RFP) count experiments in S. cerevisiae.
• Assessed the association between model predictions and RFP-derived translational slowdowns.

Main Results:

• All five evaluated models showed strong associations with RFP data, indicating potential for predicting translational slowdowns.
• A weak correlation was observed between RFP counts from independent experiments for the same genes.
• This suggests limitations in the current experimental estimation of translation rates using RFP data.

Conclusions:

• Computational models show promise for estimating translation rates and identifying slowdowns.
• The low reproducibility of RFP data raises concerns about its efficacy as a sole measure of translation rates.
• Computational models may offer a more robust approach to understanding translation dynamics.