Table 3 Description of parameter sweep

scran

Size factors normalization

{True, False }

ERCC counts normalization

{True, False }

Assay type

{logcounts, counts }

High variance genes

{ 100, 300, 500, 1000, 2000, 3000 }

Dimension of latent space

{ 2, 8, 10, 16, 32, 50, 64, 128 }

Seurat

Normalization method

{LogNormalize, CLR }

Criteria for high variance genes

{vst, mvp, dist }

High variance genes

{ 100, 300, 500, 1000, 2000, 3000 }

Dimension of latent space

{ 2, 8, 10, 16, 32, 50, 64, 128 }

ZinbWave

Gene covariates

{ True, False }

Epsilon (regularizer)

{ 200, 500, 1000, 2000 }

High variance genes

{100, 300, 500, 1000, 2000, 3000 }

Dimension of latent space

{2, 8, 10, 16, 32, 50, 64, 128 }

DCA

Dispersion and reconstruction

{zinb-conddisp, zinb, nb-conddisp, nb }

Batch normalization

{True, False }

Dimension of the latent space

{ 2, 8, 10, 16, 32, 50, 64, 128 }

Number of training epochs

{ 20, 50, 100, 200, 300, 500, 1000 }

Normalize counts

{True, False }

Scale variance

{True, False }

Log normalization

{True, False }

Dropout rate

{0, 0.1 }

Number of hidden neurons

{64, 128, 256 }

Random seed

{0, 1, 2, 3, 4 }

scVI

Number of hidden neurons

{ 64, 128, 256 }

Number of training epochs

{20, 50, 100, 200, 300, 500, 1000 }

Learning rate

{ 1e-2, 1e-3, 1e-4 }

Dropout rate

{ 0, 0.1 }

Layers

{1, 2 }

Dimension of the latent space

{ 2, 8, 10, 16, 32, 50, 64, 128 }

Dispersion

{gene, gene-cell }

Reconstruction loss

{ nb, zinb }

Random seed

{0, 1, 2, 3, 4 }