Fig. 3From: Distribution Analyzer, a methodology for identifying and clustering outlier conditions from single-cell distributions, and its application to a Nanog reporter RNAi screenDistinguishing between siRNA conditions using median cell- and distribution-based clustering. a The weighting along the first three PCs for each siNlk, siGFP and siSox2 condition (well) with >100 cells is plotted as a point in three dimensions. The mean (presumed null) effect across all conditions is shown as a grey sphere. Grey circle defines all conditions within a 2 Z-score Hellinger distance from the mean effect. Left: siNlk, siGFP and siSox2 wells are colored in red, green and yellow, respectively. Center: Color assignment is determined by 3 medoid clustering of the median cell fluorescence. Right: Color assignment is determined by 3 medoid clustering of the first 4 PCs weights. b Mean fluorescence distribution for siGFP, siSox2 and siNlk conditions (green, yellow and red, respectively) and mean distribution across all conditions in siRNA screen (grey). Distribution mean is derived from average over all replicates in distribution score (PC) space. The distributions from several representative null-effect wells are shown in black. c Median cell fluorescence for each control siRNA condition (well), ordered by category (siNlk, siGFP and siSox2). Colors represent 3 medoid clustering either by median cell fluorescence (left) or by distribution (right), as in Fig. 3a. Values beneath each siRNA represent number of conditions assigned to each clusterBack to article page