Speaker
Description
Ribosome profiling provides codon-resolution snapshots of translation, but quantitative interpretation requires accurate forward models of ribosome traffic. We revisit mRNA translation stalling by comparing the widely used mean-field Ribosome Flow Model (RFM) with a deterministic discrete-particle TASEP formulation. We found that mean-field factorization can produce artifacts, including exaggerated upstream jams near low-density/high-density transitions, artificial backfilling at clusters of slow codons, and shifted phase boundaries. To address this, we develop a deterministic credit-based hopping and drop-off framework that preserves strict exclusion at codon resolution while reproducing ensemble-average TASEP behavior without Monte Carlo sampling. The model includes ribosome-footprint constraints and site-specific drop-off, and is computationally efficient for large parameter scans in inference workflows. We classify profiles into four regimes: low density (LD), high density (HD), local high density (LHD), and local density enhancement (LDE), each linked to interpretable rate ratios. Finally, we use measurement-aware forward mappings to generate ribosome-profiling-like observations from predicted densities under finite positional resolution and count noise.
