Abstract
RNA sequencing (RNA-seq) is often used to unravel the regulation of gene expression, as it provides an in-depth view of the transcribed RNA and a relative quantification of each gene or transcript. However, several studies have found that variations in RNA transcript levels do not necessarily correlate well with protein levels [1], suggesting that translation also plays an important role. To address this need to study the impact of translation in gene expression regulation, the translatome can be studied by selecting ribosome-bound RNAs, releasing ribosomes, and then sequencing the selected population of RNAs as in RNA-seq. Polysome sequencing POL-seq is a census assay that provides relative expression levels of genes/transcripts during translation. We have developed a bioinformatics pipeline to jointly analye these transcriptome and translatome fractions, using the workflow manager Snakemake [2] and conda [3] environments. We propose a lightweight wrapper system to facilitate interoperability on different clusters or computers without the need for an Internet connection and to allow the preservation of the shell command visualization during the Snakemake dry-run. This pipeline includes steps, very similar to those used for RNA-seq, namely primary and secondary analysis where both sources of information are integrated. The primary analysis part of the pipeline consists of 4 steps: quality control of the raw data, cleaning of the reads, mapping to the reference genome and counting the mapped reads against the reference. The secondary analysis part is the main step of this workflow and concerns the statistical part. Here we have developed a filtering method. All samples are normalized together to avoid any comparison problems between fractions. Then, we combine the comparisons of each fraction, visualized by a log-log plot, and filter the data according to their transcriptional and translational expression levels. This allows us to separate mRNAs into those that are regulated by transcription only, by translation only, or by both transcription and translation, and to determine whether these regulatione have a combined or opposite effects. The differential transcripts are then enriched with gene ontologies and pathways.This pipeline was used in a published paper with our collaborators [4], we illustrate our pipeline work with figures from that paper.