Skip to main content

Table 2 Details of experimental results based on NMI scores for various dimension reduction algorithms, including the DR-A, PCA, ZIFA, scVI, SAUCIE, t-SNE, and UMAP methods. We carried out the experiments using the Rosenberg-156 k, Zheng-73 k, Zheng-68 k, Macosko-44 k, and Zeisel-3 k datasets. These dimension reduction algorithms were investigated with (a) 2 latent dimensions (K = 2), (b) 10 latent dimensions (K = 10), and (c) 20 latent dimensions (K = 20)

From: A deep adversarial variational autoencoder model for dimensionality reduction in single-cell RNA sequencing analysis

AlgorithmRosenberg-156 kZheng-73 kZheng-68 kMacosko-44 kZeisel-3 k
(a) K = 2
 DR-A0.55730.84570.59310.49360.7263
 PCA0.25230.33960.25380.29840.4721
 ZIFA0.30490.37940.28100.31200.4250
 scVI0.51990.82610.54170.45990.7006
 SAUCIE0.40460.43040.27490.27070.4622
 t-SNE0.43430.65620.40810.40910.7103
 UMAP0.55910.65070.43770.41840.7214
(b) K = 10
 DR-A0.58500.85030.57560.51560.7893
 PCA0.32760.56120.38770.42430.5559
 ZIFA0.50740.83540.51520.47850.7807
 scVI0.58210.80600.55710.51550.7606
 SAUCIE0.47730.42090.31470.28740.5110
 t-SNEN/AN/AN/AN/AN/A
 UMAP0.57350.69110.43930.41290.7413
(c) K = 20
 DR-A0.58420.80020.58880.51760.7639
 PCA0.37610.56230.38740.43060.5561
 ZIFAN/AN/AN/AN/A0.7114
 scVI0.58310.79760.56910.51050.7419
 SAUCIE0.47400.42540.29520.27750.4808
 t-SNEN/AN/AN/AN/AN/A
 UMAP0.56560.69060.44130.41770.7419
  1. N/A denotes that we could not run the given algorithm