Accurate Recovery of Ribosome Positions Reveals Slow Translation of Wobble-Pairing Codons in Yeast

Abstract

Ribosome profiling quantitatively captures ribosome locations during translation. The resulting profiles of ribosome locations are widely used to study translational speed. However, an accurate estimation of the ribosome location depends on identifying the A-site from ribosome profiling reads, a problem that was previously unsolved. Here, we propose a novel method to estimate the ribosome A-site positions from high-coverage ribosome profiling reads. Our model allows more reads to be used, accurately explains the 3-nt periodicity of ribosome profiling reads from various lengths, and recovers consistent ribosome positions across different lengths. Our recovered ribosome positions are correctly highly skewed toward a single frame within a codon. They retain subcodon resolution and enable detection of off-frame translational events, such as frameshifts. Our method improves the correlation with other estimates of codon decoding time. Furthermore, the refined profiles show that yeast wobble-pairing codons are translated slower than their synonymous Watson-Crick-pairing codons. These results provide evidence that protein synthetic rate can be tuned by codon usage bias.

Identifying A- and P-site locations on ribosome-protected mRNA fragments using Integer Programming

“…There is no independent experimental method to verify the accuracy of identified A-site locations using our method or any other method 4,5,52–55,6,8–10,12,41,42,51 . We argue that the well-established ribosome pausing at particular PPX sequence motifs is the best available means to differentiate the accuracy of existing methods.…”

“…The A-site offset tables generated using these methods for our analyzed datasets in S. cerevisiae and mESCs are presented in Supplementary Table S9. To determine the A-site profiles using the ‘ribodeblur’ method created by Wang and co-workers 6 , we ran the source code available in GitHub (https://github.com/Kingsford-Group/ribodeblur-analysis/releases/tag/v0.1) on our datasets and added a custom Python script to generate the ‘deblurred’ A-site profiles. For Rpbp 41 , the publicly available software was downloaded and run locally to obtain the A-site offsets.…”

Rationalizing Translation Elongation by Reinforcement Learning

“…Admittedly, the studies of technical bias and shortcomings [43,54] of ribosome profiling have also attracted wide interest in the computational biology community [55][56][57][58]. Without any doubt, understanding these problems would also be beneficial for deciphering the biological determinants of translation elongation dynamics.…”

Analysis of Ribosome Stalling and Translation Elongation Dynamics by Deep Learning

“…Although a large amount of sequencing data have been produced by ribosome profiling, researchers are still challenged by the complexity, heterogeneity, and insufficient coverage of these data during the data analysis process (Brar and Weissman, 2015;Ingolia, 2014Ingolia, , 2016Wang et al, 2016aWang et al, , 2016b. Recently, deep learning has become one of the most popular and powerful techniques in the machine learning field Hinton and Salakhutdinov, 2006).…”