Skip to main content
Fig. 3 | Genome Biology

Fig. 3

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

Fig. 3

Wider gRNA abundance distributions in the plasmid library increase asymmetry of before/after ratios. a–c Simulation results using three libraries with different widths of gRNA abundance distribution. a log2 count distribution of simulated libraries libA, libB, and libC. Distributions were generated as log-normal distributions with same log-mean, but differing log-sd and then size-normalized. Library distribution widths were as follows: 66.5 (libA), 17.8 (libB), and 4.8 (libC). b Reproducibility of LFC between two replicates. The simulations were performed with libA, libB, or libC. All gRNAs are shown in blue, and gRNAs without fitness effect are highlighted in gray. c Fraction of gRNAs with LFC <−1 (left) and LFC >+1 (right) for simulations with libA, libB, or libC. gRNAs used for frequency calculation do not have fitness effects. d–e Dataset from screen performed in mESCs using two different libraries lib1 and lib2 (libraries Mouse_V1 and Mouse_V2 respectively from Tzelepis et al. [27]). Library distribution widths were as follows: 8.8 (lib1) and 5.0 (lib2). d log2 count distributions of lib1 and 2, normalized to total count. e Reproducibility of LFC between two replicates in the screens performed with lib1 or lib2. All gRNAs are shown in blue. f Replicate correlation as a function of library distribution width for different published screens [2, 27, 40, 41, 43, 44]. Data for Avana, Human_V1 (also known as Score), Mouse_V1, and Mouse_V2 is highlighted in color

Back to article page