Table 1.
Method | Distribution assumption | Designed for | Test statistic | Run time | Version [Ref.] | Input |
---|---|---|---|---|---|---|
BPSC | Beta-Poisson | Single cell | z-test | Hours | 0.99.2 (Vu et al., 2016) | CPM |
DEsingle | Zero-Inflated Negative Binomial | Single cell | Likelihood ratio test | Hours | 1.2.1 (Miao et al., 2018) | raw counts |
MAST | Normal (Generalized linear hurdle) | Single cell | Likelihood ratio test | Minutes | 1.8.2 (Finak et al., 2015) | log2(CPM+1) |
monocle | Normal (Generalized additive model) | Single cell | Likelihood ratio test | Minutes | 2.10.1(Qiu et al., 2017) | raw counts |
DESeq2 | Negative Binomial | Bulk cell | Wald test | Minutes | 1.22.2 (Love et al., 2014) | raw counts |
edgeR | Negative Binomial | Bulk cell | Quasi-likelihood F-test | Minutes | 3.24.3 (Robinson et al., 2010) | raw counts |
limmatrend | Normal (linear model) | Bulk cell | Empirical-Bayes Moderated t-statistics | Seconds | (Law et al., 2014) | log2(CPM+1) |
t-test | Normal | General | t-test | Seconds | (Welch, 1947) | log2(CPM+1) |
Wilcoxon | Nonparametric | General | Wilcoxon | Minutes | (Hollander et al., 2013) | log2(CPM+1) |