Fig. 2
From: scReClassify: post hoc cell type classification of single-cell rNA-seq data
Evaluation of scReClassify on simulated datasets. In (a) and (b), each cell is shown as a solid point on each tSNE plot and coloured coded by its true cell type. The initial cell type labels (left panels in A and B) are the true cell types but with either 10% (a) or 20% (b) mislabelling. The solid point with an open ring represent incorrectly label cells and the ring colour represent the mislabelled type. In (c) and (d), scRNA-seq with varying percentages of mislabelled cells were simulated. Baselines represent the (c) mean accuracy and (d) adjusted Rand index (ARI) calculated from the ground truth and the initial annotations. Performance of scReClassify are shown as boxes coloured according to the percentages of mislabelled cells