Table 3 Performance of GRouNdGAN in generating realistic scRNA-seq data using the Dahlin, Tumor-All, and Tumor-malignant datasets

From: GRouNdGAN: GRN-guided simulation of single-cell RNA-seq data using causal generative adversarial networks

 

Dahlin

Tumor-All

Tumor-malignant

Training

Testing

Training

Testing

Training

Testing

Cosine distance

0.00011

0.00012

0.00141

0.00061

0.00059

0.00098

Euclidean distance

51

53

227

145

134

174

MMD

0.014

0.026

0.020

0.021

0.014

0.015

RF AUROC

0.52

0.54

0.53

0.52

0.53

0.57

miLISI

1.91

1.91

1.90

1.89

1.90

1.89

  1. Each gene in the imposed GRN of GRouNdGAN is regulated by 15 TFs (constructed using GRNBoost2 from the experimental training set). For the first three metrics, a value closer to zero is preferred, for RF AUROC a value closer to 0.5 is preferred, and for miLISI a value closer to 2 is preferred. The metrics are calculated between a simulated dataset of 1000 cells and a set of 1000 real cells (all cells in the test set and 1000 randomly selected cells from the training set). For the first two metrics, the values correspond to the distance of the mean centroids of the real and simulated cells.
Back to article page