Fig. 7: Prediction of wild-type RA- and RA+ cells using the validated hierarchal random forest model. | Communications Biology

Fig. 7: Prediction of wild-type RA- and RA+ cells using the validated hierarchal random forest model.

From: Computational profiling of hiPSC-derived heart organoids reveals chamber defects associated with NKX2-5 deficiency

Fig. 7

a UMAP projections of RA- and RA+ single cells at day 30 grouped by cell type annotation. b UMAP projections of wild-type RA- and RA+ single cells at day 30 with Anomalies removed grouped by (i) cell types, (ii) treatment, and (iii) predicted cell types and zone. c Sankey diagram of prediction results of wild-type RA- and RA+ cells at day 30 (Anomalies removed). Table provides cell type and conditional zone accuracies. d Expression levels of genes used to make prediction decisions that either correlate with the predictions (A and V) or experimental conditions (RA- and RA+). ****p ≤ 0.0001, Bonferroni corrected p values from Wilcox test, exact p-values and sample numbers available in a Source data file. N = 308 A_RA-, 580 V_RA-, 947 A_RA+, 195 V_RA+. e Bar plot of side predictions in WT cells at day 30.

Back to article page