ARMOR: an Automated Reproducible MOdular workflow for preprocessing and differential analysis of RNA-seq data

Abstract

7 1 These authors contributed equally 8 2 The order of the shared first authors was determined randomly, using the sample() function in R v3.5.2, 9with the random seed 1552397284.Abstract 15 The extensive generation of RNA sequencing (RNA-seq) data in the last decade has resulted in 16 a myriad of specialized software for its analysis. Each software module typically targets a specific 17 step within the analysis pipeline, making it necessary to join several of them to get a single cohesive 18 workflow. Multiple software programs automating this procedure have been proposed, but often lack 19 modularity, transparency or flexibility. We present ARMOR, which performs an end-to-end RNA-seq 20 data analysis, from raw read files, via quality checks, alignment and quantification, to differential ex-21 pression testing, geneset analysis and browser-based exploration of the data. ARMOR is implemented 22 using the Snakemake workflow management system and leverages conda environments; Bioconduc-23 tor objects are generated to facilitate downstream analysis, ensuring seamless integration with many 24 R packages. The workflow is easily implemented by cloning the GitHub repository, replacing the sup-25 plied input and reference files and editing a configuration file. Although we have selected the tools 26 currently included in ARMOR, the setup is modular and alternative tools can be easily integrated. 27 1 Introduction 28 Since the first high-throughput RNA-seq experiments about a decade ago, there has been a tremendous 29 development in the understanding of the characteristic features of the collected data, as well as the meth-30 ods used for the analysis. Today there are vetted, well-established algorithms and tools available for 31 many aspects of RNA-seq data analysis (Conesa et al. 2016; Van Den Berge et al. 2018). In this study, 32 we capitalize on this knowledge and present a modular, light-weight RNA-seq workflow covering the 33 most common parts of a typical end-to-end RNA-seq data analysis. The application is implemented 34 using the Snakemake workflow management system (Köster and Rahmann 2012), and allows the user 35 to easily perform quality assessment, adapter trimming, genome alignment, transcript and gene abun-36 dance quantification, differential expression analysis and geneset analyses with a simple command, after 37 specifying the required reference files and information about the experimental design in a configuration 38 file. Reproducibility is ensured via the use of conda environments, and all relevant log files are retained 39 for transparency. The output is provided in state-of-the-art R/Bioconductor classes, ensuring interoper-40 ability with a broad range of Bioconductor packages. In particular, we provide a template to facilitate 41 browser-based interactive visualization of the quantified abundances and the results of the statistical 42 analyses with iSEE (Rue-Albrecht et al. 2018).43 Among already existing pipelines for automated reference-based RNA-seq analysis, several focus 44 either on the p...

An atypical class of non-coding small RNAs produced in rice leaves upon bacterial infection

“…GSE67588). Pre-processing and downstream statistical analysis of the mRNA-seq data was performed using the ARMOR workflow (33) to compute differential expression metrics for genes annotated in the MSU7 Nipponbare annotation in comparisons involving samples inoculated with a virulent strain versus samples inoculated with the corresponding T3SS mutant strain or mock buffer (i.e. BLS256vsMOCK, BAI3vsH2O, MAI1vsMOCK, BAI11vsMOCK, BAI3vsBAI3H).…”

“…For the analysis of differential gene expression using mRNA-seq data, we applied the ARMOR pipeline (33) with the following parameter values specified in the config file: additional_salmon_index: “-k 31”, additional_salmon_quant: “--seqBias –gcBias”. Note that whenever available in the sample metadata information, a ‘experiment’ factor term was included in the statistical model for sRNA loci and gene DE analysis in order to block for experimental batch effects.…”

Global analysis of human SARS-CoV-2 infection and host-virus interaction

“…The first one, based on STAR (Dobin et al, 2013), carries out alignments to the human genome, allowing for recognition of reads mapping to introns which could then be relevant for alternative splicing and regulatory analyses. The second one is based on ARMOR (Orjuela, Huang, Hembach, Robinson, & Soneson, 2019), aligning the reads to the transcriptome instead.…”

“…The SRA files were transformed to fastq files prior to applying the ARMOR workflow (Orjuela et al, 2019). Snakemake is applied to facilitate a reproducible and robust computational workflow.…”

“…The RNA-sequencing reads from a COPD related dataset involving human clinical samples were downloaded from the GEO database (GSE57148). These reads were then analyzed using pipeline 2 from project 1 as described above (Orjuela et al, 2019). Differentially expressed genes from the case vs. control comparison were then subjected to the same signaling pathway and functional enrichment analyses produced from the second pipeline in Project 1.…”