Skip to main content
Fig. 3 | Genome Biology

Fig. 3

From: Vireo: Bayesian demultiplexing of pooled single-cell RNA-seq data without genotype reference

Fig. 3

Evaluation of Vireo on data from multiplexed human PBMCs. ac Results obtained on two datasets consisting of two pools of 4 samples each, as well as a third dataset consisting of the union of all 8 samples. a Concordance of singlet assignment and doublet detection between Vireo without genotype data and Demuxlet applied with complete genotype reference. Bars denote the number of cells assigned to each sample, either considering cells that were consistently assigned by both Vireo and Demuxlet (blue), or assigned exclusively by Vireo (green) or Demuxlet (red). b Concordance of doublet detection between Vireo and Demuxlet when varying the assignment threshold for each method. Note, p denotes the threshold prob_doublet in Vireo (x-axis) and Demuxlet (y-axis) respectively, and n denotes the number of detected doublets. Assignment of cells in a is based on the most probable sample assignment, considering all cells that were not detected as doublet. Cells with a doublet probability (p_doublet >0.9 in Vireo; >2/3 in Demuxlet) were labelled as doublet cells and are considered in b. c Alignment of samples, when applying Vireo separately to the three datasets considered in a. Values in the heatmap denote the fraction of concordant genotype states between pairs of samples from both Vireo runs, considering variants with a read coverage of at least 10 UMIs per sample. df Results from a second experiment, consisting of two datasets with the same 8 samples pooled in two different conditions: unstimulated and stimulated. Results shown correspond to the panels in ac

Back to article page