Scatter Plot from Scientific Research

Overlap between a reference k -NN graph (inferred using all transformations on deeply sequenced data and taking the intersection) and the k -NN inferred on data downsampled to match typical 10x data (5,000 counts per cell) for five datasets with five replicates each. To compare and aggregate results across the different datasets, Relative overlap ( a – c ), which was computed by dividing, for each dataset, the overlap by its average across all transformations, fixing k = 50 and using a dataset-specific number of PCA dimensions (Extended Data Fig. 7 shows the underlying, unaggregated data).

DOI