Skip to main content
. 2018 Dec 24;35(16):2865–2867. doi: 10.1093/bioinformatics/bty1044

Fig. 1.

Fig. 1.

Comparison of feature selection methods. (A and B) Accuracy in identifying DE genes in simulated data. (C and D) Reproducibility of features across five mouse embryo and four human pancreas datasets. (E and F) Average fold-change in expression of reproducible features. (C–F) Each point represents a pair of datasets and the horizontal lines indicate the mean across all pairs. PCA scored genes by their loadings for the top components, Gini is the method used by GiniClust (Jiang et al., 2016), Cons is the consensus across all other methods

OSZAR »