Fig. 3: Lack of whole-cell negative correlation between Xist and Tsix argues against mechanisms involving transcript recruitment.
a An example image of data showing individual molecules of mature Xist (green) and Tsix (magenta) surrounded by an orange box. This is the same cell as shown in Fig. 2d. b Conceptual diagram showing the expected data. Recruitment of transcription regulators on a whole-cell level would result in negative correlation when investigating whole-cell transcript levels (left). Otherwise, there would be no whole-cell correlation (right). c The joint probability distribution of whole-cell levels of Xist and Tsix mature transcripts in three different replica experiments. The correlation coefficient (rho) and the significance (p) of correlations are determined by the Spearman correlation test, and no significant correlation is found in two out of three replicas. d The joint probability distribution of the randomized data, which used the same data but randomly permuted the single-cell relationship between the two genes. A two-dimensional two sided KS-test56 (p-value shown) determined no significant difference between the original data and the randomized version in all replicas. Xist data is binned by 20 transcripts. Tsix data is binned by 1 transcript due to its low expression. Color code corresponds to the negative logarithm base 10 of the probability. Replica 1, n = 963 cells, 2289 Tsix transcripts, 5226 Xist transcripts; replica 2, n = 1063 cells, 3322 Tsix transcripts, 5690 Xist transcripts; replica 3, n = 814 cells, 3116 Tsix transcripts, 19701 Xist transcripts.
