Understanding biases in ribosome profiling experiments reveals signatures of translation dynamics

All ribosome profiling experiments analyzed involve attaching a known sequence to the 3′ end of RNA footprints to which a reverse transcription primer can be annealed. Some experiments use polyA tailing for this purpose, while others attach an oligonucleotide linker sequence. For experiments using polyA tailing, reads were trimmed from the end back to the first base that wasn’t an A or an N. For experiments using linker sequences, linkers were located in reads by local alignment with the expected sequence and trimmed. Trimmed reads were first mapped to yeast rRNA sequences with bowtie2[16], and any reads that mapped were filtered out. Remaining reads were mapped with tophat2[15] to the yeast genome (version EF4) and spliced transcriptome (using transcript models from the Saccharomyces Genome Database’s .gff dated Fri Apr 11 19:50:03 2014). Unmapped reads had any terminal stretches of A trimmed and were put through tophat2 again to recover potential mappings overlapping transcript polyA tails, although this has minimal impact on the analysis presented here. The reverse transcription process used to convert footprints to DNA can add untemplated bases to the end of intermediate antisense DNA products, which ultimately end up located at the beginning of sequencing reads [12]. We observed that the rate at which this happens varies considerably between different experiments. To prevent these untemplated bases from potentially shifting the