seq2ribo: Structure-aware integration of machine learning and simulation to predict ribosome location profiles from...

Area of Science:

• Computational Biology
• Molecular Biology
• Bioinformatics

Background:

• Ribosome dynamics are crucial for protein expression, but current methods like ribosome profiling (Ribo-seq) require genomic context, limiting applications in synthetic biology.
• Existing simulation models, such as the Totally Asymmetric Simple Exclusion Process (TASEP), oversimplify translation by focusing only on codon elongation times.

Purpose of the Study:

• To develop a novel computational framework, seq2ribo, for predicting ribosome A-site locations using only mRNA sequence as input.
• To enable de novo sequence design for applications like messenger RNA (mRNA) vaccines by overcoming limitations of existing methods.

Main Methods:

• Implemented a hybrid simulation and machine learning approach combining a structure-aware TASEP (sTASEP) with a refinement polisher model.
• sTASEP incorporates codon wait times and structural features (local angles, base-pairing, positional buckets) to model translation.
• The polisher model refines ribosome distributions simulated by sTASEP.

Main Results:

• seq2ribo achieves high-fidelity ribosome location predictions across diverse cell types, outperforming existing baselines.
• Demonstrated significant reductions in transcript-level error (up to 35.8%) and structural errors (43.3%-97.3%) compared to sequence-only methods.
• Achieved high correlations with experimental translation efficiency (up to 0.795 Spearman) and protein expression (0.689 Spearman).

Conclusions:

• seq2ribo offers a powerful, sequence-only tool for predicting ribosome dynamics, advancing synthetic biology and rational mRNA sequence design.
• The framework eliminates the need for expression-level data or genomic context, facilitating the optimization of mRNA sequences for novel applications.