Fig. 1

Illustrated representation of the algorithms workflow. A A simple 2-dimensional example, based on a dataset containing two markers with 4 populations, 2 of which are connected, is clustered in 2 < m < k populations. Each clustering result is inspected by assessment of marker distribution through the dip test and marker spread test. B Result of metaclustering of the example dataset, split in 2, 4 or 8 populations. C Marker distributions in the formed metaclusters. Green density plots pass the dip test and marker spread test, red distributions denote failed markers due to non-unimodal distribution. D For every metaclustering result (denoted with i), all marker distributions for the m number of metaclusters are taken together and the fraction of passed distributions is taken. E Representative diagnostic plot for a larger dataset. The values of the Unimodality set \({U}_{i}\) are plotted on the y-axis versus the number of metaclusters assessed on the x-axis in red. A sigmoidal curve (blue) is fitted to this data. A plateau is reached where the fraction of unimodal distributions scarcely increases with increasing numbers of metaclusters