Extended Data Fig. 8: Mixscape analysis. | Nature

Extended Data Fig. 8: Mixscape analysis.

From: Precisely defining disease variant effects in CRISPR-edited single cells

Extended Data Fig. 8: Mixscape analysis.

(a) Heatmap comparing ground truth genotype dosage (x-axis) and Mixscape prediction assignment of a cell’s editing status (NP: non-perturbed, KO: knock-out). This panel predicts FBXO11 editing status in Daudi cells, where CRISPR vs control status was input into the Mixscape model (genotyping blind). Cells below the dotted line refer to predictions made by Mixscape, while cells above the dotted line were input into the Mixscape model as controls. Green reflects correct assignments by Mixscape and red reflects incorrect assignments. Grey reflects either assignments directly supplied to the model rather than predicted by it or, for dosage of 1, assignments that are not possible to define as accurate (i.e. if the biological interpretation of dosage of 1 is better characterized as NP or KO). (b) Same as (a), except genotyping information of Dosage 0 versus Dosage 1 or 2 was input into the Mixscape model. (c) Same as (a), except genotyping information of Dosage 0 or 1 versus Dosage 2 was input into the Mixscape model. (d) Scatterplots showing concordance of differentially expressed gene Z-scores, as calculated from our linear model using sequenced genotype (x-axis) and from an otherwise identical model where cell editing status has been assigned by Mixscape. Each plot refers to different inputs to the Mixscape model, analogous to (A-C). The solid line is the identity line. (e) Barplots displaying percentage of genes identified through differential expression of Mixscape results that are concordant with the results from our linear models using sequenced dosage (pink) and true cell editing status (brown). X-axis refers to the three different inputs supplied to Mixscape. (f-h) Same as (a-c) but for prediction of CD45 editing status in primary T cells. (i) Same as (d) for CD45. Only scatter plots corresponding to (F) and (H) are present due to Mixscape being unable to identify any KO cells in one of the inputs to Mixscape. (j) Same as (e) for CD45. Only bars corresponding to (f) and (h) are present.

Back to article page