Table 1 Description of the 14 batch-effect correction methods
From: A benchmark of batch-effect correction methods for single-cell RNA sequencing data
Tools | Programming language | Batch-effect-corrected output | Methods | Reference package version |
|---|---|---|---|---|
Seurat 2 (CCA, MultiCCA) | R | Normalized canonical components (CCs) | Canonical correlation analysis and dynamic time warping | Butler et al. [4], Seurat package version 2.3.4 |
Seurat 3 (Integration) | R | Normalized gene expression matrix | Canonical correlation analysis and mutual nearest neighbors-anchors | Stuart et al. [12], Seurat package version 3.0.1 |
Harmony | R | Normalized feature reduction vectors (Harmony) | Iterative clustering in dimensionally reduced space | Korsunsky et al. [13], Harmony version 0.99.9 |
MNN Correct | R | Normalized gene expression matrix | Mutual nearest neighbor in gene expression space | Haghverdi et al. [5], Scran package version 1.12.0 |
fastMNN | R | Normalized principal components | Mutual nearest neighbor in dimensionally reduced space | Haghverdi et al. [5], Lun ATL [7], Scran package version 1.12.0 |
ComBat | R | Normalized gene expression matrix | Adjusts for known batches using an empirical Bayesian framework | Johnson et al. [1] |
limma | R | Normalized gene expression matrix | Linear model/empirical Bayes model | Smyth et al. [2], limma version 3.38.3 |
scGen | Python | Normalized gene expression matrix | Variational auto-encoders neural network model and latent space | Lotfollahi et al. [16], 2019, scGen version 1.0.0 |
Scanorama | Python/R | Normalized gene expression matrix | Mutual nearest neighbor and panoramic stitching | Hie et al. [9], Scanorama version 1.4. |
MND-ResNet | Python | Normalized principal components | Residual neural network for calibration | Shaham et al. [15] updated code to Python 3 |
ZINB-WaVE | R | Normalized feature reduction vectors (ZINB-WaVE)/normalized gene expression matrix | Zero-inflated negative binomial model, extension of RUV model | Risso et al. [6], ZINB-WaVE version 1.6.0 |
scMerge | R | Normalized gene expression matrix | Stably expressed genes (scSEGs) and RUVIII model | Lin et al. [18], scMerge version 1.1.3 |
LIGER | R | Normalized feature reduction vectors (LIGER) | Integrative non-negative matrix factorization (iNMF) and joint clustering + quantile alignment | Welch et al. [14], liger version 1.0 |
BBKNN | Python/R | Connectivity graph and normalized dimension reduction vectors (UMAP) | Batch balanced k-nearest neighbors | Polański et al. [10], bioRxiv. BBKNN version 1.3.2 |