gscreend: modelling asymmetric count ratios in CRISPR screens to decrease experiment size and improve phenotype detection

• Published in: Genome Biology Vol. 21; no. 1; p. 53
• Main Authors: Imkeller, Katharina, Ambrosi, Giulia, Boutros, Michael, Huber, Wolfgang
• Format: Journal Article
• Language: English
• Published: England BioMed Central 02.03.2020; BMC
• Subjects: Animals; Asymmetry; Biology-based model; Cell Proliferation; Clustered Regularly Interspaced Short Palindromic Repeats; CRISPR; CRISPR screen; Deoxyribonucleic acid; DNA; Enzymes; Experimental design; Experiments; Gene expression; Gene-Environment Interaction; Generative probabilistic model; Genetic perturbation screen; Genome-Wide Association Study - methods; Genomes; Genomics - methods; Genotype; Genotype & phenotype; gscreend package; HCT116 Cells; HEK293 Cells; Humans; Method; Mice; Models, Genetic; Phenotypes; Power; RNA, Guide, CRISPR-Cas Systems - chemistry; RNA, Guide, CRISPR-Cas Systems - genetics; Simulation; Software; Biology-based model; Bioconductor; gscreend package; Generative probabilistic model; Experimental design; CRISPR screen; Genetic perturbation screen
• ISSN: 1474-760X; 1474-7596; 1474-760X
• DOI: 10.1186/s13059-020-1939-1

Pooled CRISPR screens are a powerful tool to probe genotype-phenotype relationships at genome-wide scale. However, criteria for optimal design are missing, and it remains unclear how experimental parameters affect results. Here, we report that random decreases in gRNA abundance are more likely than increases due to bottle-neck effects during the cell proliferation phase. Failure to consider this asymmetry leads to loss of detection power. We provide a new statistical test that addresses this problem and improves hit detection at reduced experiment size. The method is implemented in the R package gscreend, which is available at http://bioconductor.org/packages/gscreend.