shRNA-seq data analysis with edgeR

Abstract

Pooled short hairpin RNA sequencing (shRNA-seq) screens are becoming increasingly popular in functional genomics research, and there is a need to establish optimal analysis tools to handle such data. Our open-source shRNA processing pipeline in edgeR provides a complete analysis solution for shRNA-seq screen data, that begins with the raw sequence reads and ends with a ranked lists of candidate shRNAs for downstream biological validation. We first summarize the raw data contained in a fastq file into a matrix of counts (samples in the columns, hairpins in the rows) with options for allowing mismatches and small shifts in hairpin position. Diagnostic plots, normalization and differential representation analysis can then be performed using established methods to prioritize results in a statistically rigorous way, with the choice of either the classic exact testing methodology or a generalized linear modelling that can handle complex experimental designs. A detailed users’ guide that demonstrates how to analyze screen data in edgeR along with a point-and-click implementation of this workflow in Galaxy are also provided. The edgeR package is freely available from http://www.bioconductor.org.

Genome-wide CRISPR-Cas9 Screens Reveal Loss of Redundancy between PKMYT1 and WEE1 in Glioblastoma Stem-like Cells

“…While edgeR has been mainly used for examining changes in steady state mRNA levels from SAGE and RNA-seq data, by design, edgeR was intended for use with any type of count based sequence tag data in complex libraries, including nucleic acid bar codes (Dai et al, 2014). To do so, edgeR models the count variance across replicates as a nonlinear function of the mean counts using a negative binomial distribution while accounting for over all data dispersion.…”

Coordinated Splicing of Regulatory Detained Introns within Oncogenic Transcripts Creates an Exploitable Vulnerability in Malignant Glioma

“…Non-normalized count numbers per shRNA were calculated by aligning the sequencing reads to the input shRNA library allowing for a single mismatch. The counts tables were then used as an input for the edgeR based shRNA-seq bioinformatical tool for pooled shRNA screen analysis (Dai et al, 2014). …”

MDM4 is an essential disease driver targeted by 1q gain in Burkitt lymphoma

“…Cells were harvested on day 2 (one replicate) or day 14 (2 replicates) post-transduction. p-values for shRNA depletion were calculated with the edgeR package (Dai et al, 2014) and shRNA p-values were collapsed into gene scores using weighted Ztransformation (Kim and Tan, 2012). For comparison of differential shRNA effects between two groups, log2-transformed shRNA fold-changes were scaled with peak median absolute deviation (PMAD) normalization using the GenePattern module NormLines (Cheung et al, 2011;Reich et al, 2006).…”

“…On average 24% of shRNAs were depleted at least two-fold and shRNAs targeting core essential complexes, including the ribosome and the proteasome, were specifically lost (68% and 47%, respectively) ( Figure 1B). To evaluate the viability effect of individual gene knock-downs, we calculated weighted z-scores that combine the effect of shRNAs targeting the same gene and emphasize strong fold-changes (Dai et al, 2014;Kim and Tan, 2012) (see Methods). Common essential genes, as defined on the basis of previous RNAi screens (Hart et al, 2014), showed significantly lower scores compared to non-essential genes (p<0.001, Figure 1C).…”