Fig. 3.

Noisiness. a Assume the observation on the left panel. When this observation is obtained as multiple signals from one point, as indicated in the middle panel, the deviation of dots from the cross should be interpreted as random noise. When the observation is obtained as multiple samples from a population with heterogeneity in terms of this measurement, the variation of dots is the most important information. The population’s heterogeneity is represented by contour lines in the top right panel. b Example of source of noisiness. The bottom left panel shows the chronological periodic change of the expression level of two genes, X and Y. The right bottom panel is a two-dimensional display of the positions of (X, Y). When a particular cell type is studied and many cells of this type of cycle are evaluated for X and Y, individually (single-cell expression analysis), their coplot of (X, Y) will appear circular. This is not noise, but rather an important feature of this type of cell. However, if these value sets of X and Y are observed without chronological order, this variation can be considered as noise without meaningful explanation