Detecting ribosome collisions with differential rRNA fragment analysis in ribosome profiling data.

It has become clear in recent years that ribosomes regularly stall during translation. Such translation impairment has many causes, including exposure to ribotoxic stress agents, the presence of specific RNA structures or sequences, or a shortage of amino acids or translation factors. If they are not resolved, stalled ribosomes can lead to ribosome collisions that are continuously surveilled by various sensor proteins. This in turn initiates a cascade of signalling events that can change the physiology and behaviour of cells. However, measuring changes in collision abundance has proved challenging, and as a result, the importance of collision-mediated biological responses is still unclear. Here, we show that computational analyses of standard ribosome profiling (Ribo-seq) data enable the prediction of changes in ribosome collisions between conditions. This is achieved by using the known 3D structure of collided ribosomes to define the ribosomal RNA (rRNA) positions that are differentially digested by RNases during the Ribo-seq protocol. Comparison of the relative rRNA reads at these positions allows the relative quantification of collisions between samples, an approach we call differential ribosome collisions by Analysis of rRNA Fragments (dricARF). When applied to public datasets across multiple organisms, our approach detects changes in collision events with unprecedented accuracy and sensitivity. In addition to providing supplementary evidence for ribosome collisions, our tool has the potential to uncover novel biological processes that are mediated by them. dricARF is available as part of the ARF R package and can be accessed through https://github.com/fallerlab/ARF.