Fig. 5
From: hadge: a comprehensive pipeline for donor deconvolution in single-cell studies
Benchmarking performance. A Hadge genetic and hashing demultiplexing pipelines were benchmarked against demuxafy and cellhashR. The benchmark was performed on three samples for each pairwise comparison, for a total of four samples (mpxMS:gx12, mpxMS:gx38, demuxafy dataset, CR-438–21 dataset). B Results of hadge genetic on the onek1k cohort; each boxplot represents the distribution of percentage singlets identified across 75 pools by each genetic deconvolution tool. C Percentage of correctly matched singlets for each tool; each boxplot represents the distribution across 75 pools. D Dotplot showing the effect of the number of cells per pool on the percentage of recovered and matched singlets. The regression line represents the fit of a linear model on the percentage of singlets identified by Freemuxlet (R.2 0.35, p.adj < 0.0001)