Journal: Nucleic Acids Research

Article Title: Recovering false negatives in CRISPR fitness screens with JLOE

doi: 10.1093/nar/gkad046

Figure Lengend Snippet: False negative genes in essential protein complexes. Binary essentiality calls of essential protein complexes among colorectal cancer cell lines in the Avana data based on BAGEL derived essentiality scores. Dark blue color indicates essentiality (BF ≥ 10) and white color indicates non-essentiality (BF < 10). ( A ) Binary essentiality calls of members of 26S proteosome complex. ( B ) Binary essentiality calls of members of COP9 signalosome complex among colorectal cancer cell lines in the Avana data based on BAGEL derived essentiality scores. The relationship between screen quality ( F -measure) and the number of non- essential hits within essential protein complexes. Boxplots showing the distribution of F -measures of screens versus the number of times a non-essential call was made among members of proteosome complex ( C ) and COPS9 signalosome complex ( D ). MCM complex is an example of a CRISPR-library specific false negative. While some members of the MCM complex cannot be captured as essential in multiple colorectal screens in the Avana dataset ( E ), the complex shows more uniform essentiality in Sanger colorectal cell lines ( F ).

Article Snippet: The raw read count file of genome-wide CRISPR pooled library screens for 769 cell lines using Avana library (Broad DepMap, 2020b; Meyers et al. , 2017) (Broad Institute's DepMap Project 20Q2 release) was downloaded from the data depository at https://depmap.org/portal/ .

Techniques: Derivative Assay, CRISPR

Journal: Nucleic Acids Research

Article Title: Recovering false negatives in CRISPR fitness screens with JLOE

doi: 10.1093/nar/gkad046

Figure Lengend Snippet: JLOE approach identifies more essential genes in Avana genome-wide CRISPR-Cas9 knockout screens compared to saturation modeling approach. ( A ) Schematic description of JLOE method. ( B ) Comparison of the frequency of essential gene hits in saturation modeling versus JLOE. ( C ) Comparison of violin plots showing the distribution of the essentiality scores (Bayes Factor, BFs) of the hits in (A), in the screens where they were not observed as essential. JLOE can rescue the false negatives with a less significant increase in false positives compared to assigning binary calls using a lower BF threshold. ( D ) The number of essential gene hits using different thresholds (binary calls with BF 10, binary calls with BF 5 and mean binary call of 1 with JLOE across 100 iterations. ( E ) The number of false positive genes detected across colorectal cancer cell lines in the Avana data using same thresholds in (D). P -value annotation from t -test ind. samples with Bonferroni correction legend as follows; ns: 0.05 < P , *: 0.01 < P ≤ 0.05, **: 0.001 < P ≤ 0.01, ***: 0.0001 < P ≤ 0.001.

Article Snippet: The raw read count file of genome-wide CRISPR pooled library screens for 769 cell lines using Avana library (Broad DepMap, 2020b; Meyers et al. , 2017) (Broad Institute's DepMap Project 20Q2 release) was downloaded from the data depository at https://depmap.org/portal/ .

Techniques: Genome Wide, CRISPR, Knock-Out, Comparison