
ScreenR: analysis of pooled shRNA and CRISPR-Cas9 screens
Source:R/ScreenR-package.R
ScreenR-package.RdScreenR finds candidate hits in loss-of-function pooled shRNA and CRISPR-Cas9 screens, comparing a treatment with a control or the time points of a dropout screen. It combines the statistical models of edgeR and limma with the simplicity of the tidyverse, and provides plots for each step of the analysis.
Details
The analysis starts from a barcode count table and an annotation table,
which are stored in a ScreenR object (create_screenr_object).
The counts are then normalized (normalize_data), checked
with quality control plots, and the hits are found with several
statistical methods:
the Z-score of the barcode log2 fold changes with a gene-level permutation test (
find_gene_zscore_hit, Z-score);ROAST and FRY (
find_roast_hit);a time-course model (
find_time_hit) and a test for an effect at every time point (find_consistent_hit).
CRISPR-Cas9 screens can be checked with essential and non-essential
reference genes (compute_screen_quality) and corrected for
the copy-number bias with CRISPRcleanR
(correct_crispr_bias).
The results of the methods can be combined with
summarise_hits and further filtered with the variance and
slope filters (filter_by_variance,
filter_by_slope).
Author
Emanuel Michele Soda emanuelsoda@gmail.com