Figure 1.

The ClonoCluster method integrates transcriptome and clone clustering modalities in a tunable manner using the α parameter

(A) Schematic depicting a generic approach to single-cell barcoding that yields output clusterable by two modalities of data, transcriptome clustering, and clone clustering by recovered barcode.

(B) Schematic depicting the integration of these clone and transcriptome clustering modalities using the ClonoCluster method, in which the transcriptome nearest neighbor network graph edge weights are modified to incorporate clone clusters with a tunable free parameter, α. At α = 0, clustering is identical to traditional transcriptome clusters. At α = 1, clusters are consistent with clone barcode assignments.

(C) Sankey diagram depicting the reorganization of cell clusters from the 15 largest clone clusters present at day 2 in an in vitro hematopoiesis assay¹² with increasing α value colored by initial transcriptome cluster (top) and clone (bottom). Nodes/boxes represent clusters and ribbons depict the flow of cells between clusters.

(D) Representative plot of high-dose BRAF inhibitor-treated melanoma clonal cell line (clonal WM989 cells treated with 1 μM of the BRAF inhibitor vemurafenib¹¹) showing that cluster number approaches unique clone barcode number with increasing α value at fixed community detection resolution. “High α” values were chosen based on the approximate shoulder of the curve after which clusters rapidly break into individual clone clusters with increasing α. “Low α” values were determined by the half “high α” value rounded to the nearest 10th value. At α = 0, clustering is identical to traditional transcriptome clusters. At α = 1, clusters are consistent with clone barcode assignments.

(E) Representative Sankey diagram of the 15 largest clone clusters from the high-dose BRAF inhibitor dataset treated melanoma clonal cell line depicting rearrangements of clusters at transcriptome, low α, and high α levels, colored by initial transcriptome cluster assignment (left) and high α cluster assignment (right).