Extended Data Fig. 1: In vivo GOF CRISPRa screens identified boosters that enhance CAR-NK anti-tumor function.
From: OR7A10 GPCR engineering boosts CAR-NK therapy against solid tumours
a, Schematics of the two in vivo GOF screens that identified functional boosters that enhance CAR-NK anti-tumor function. Top, Schematics of the primary screen (1st screen) - a CRISPRa screen using SAM sgRNA library with CAR-NK92 cells in HT29 tumor model. Bottom, Schematics of the secondary screen (2nd screen) - a barcoded UMI ORF mini-screen using top hits from the primary screen with human primary CAR-NK cells in HT29 tumor model. b, Heatmap of the correlation between CRISPRa screen samples. Sample correlation is presented as the Spearman rho values. c, Plot of the multidimensional scaling of data from CRISPRa screen samples. The samples were clustered by k-means, and a convex hull was drawn around the clustered samples. d, Schematic for a Screen Analysis Method with empirical Bayes estimates for Aggregated gene scoring (SAMBA). e, Summary information of the datasets and gene sets used to benchmark CRISPR screen analysis methods. f, Schematic describing the strategy for benchmarking CRISPR screen analysis methods against six other algorithms for enrichment and depletion analyses of datasets with simulated sparsity levels. g, Line plots showing the performance characteristics of different analysis methods, when guide detection is decreased among screen samples. Overall detection performance was measured by area-under-curve (AUC) values, specificity by the partial-AUC (pAUC) with a true-positive rate > 90%, and sensitivity by the pAUC with a false-positive rate <10%. Trendlines are the loess curves (opaque line) of AUC/pAUC values at each data sparsity level, shown with the standard error +/− the mean (shaded line) (see Methods for more details). The trendline of the area-under-curve (AUC) and partial-AUC (pAUC) values for 71 screens at different levels of data sparsity (see Methods for details). Each method is presented with a loess curve (opaque line), shown with the standard error +/− the mean (shaded line). The schematic in a was created using BioRender (https://biorender.com).
