We identified cell clones by performing graph-based clustering on the square-root-transformed allele frequency matrix by first creating a neighbor graph using the FindNeighbors function in Seurat with annoy.metric = ‘cosine’ to use the cosine distance to define nearest neighbors and with k = 10. We then performed community detection using the Smart Local Moving algorithm62 using the shared nearest-neighbor graph computed using Seurat. This was implemented in the FindClonotypes function in Signac.
