From: MCMSeq: Bayesian hierarchical modeling of clustered and repeated measures RNA sequencing experiments
Method | R-Package | Description | Reference |
---|---|---|---|
MCMSeq | mcmseq | Bayesian negative binomial hierarchical model, assuming the following | Â |
 |  | hyper-parameters: T=72I4, where I4 is a 4 x 4 identity matrix, U= |  |
 |  | V=0.01, Ag= centered on empirical prior as described in 2.3, \(B=4\cdot \sigma ^{2}_{re}\) |  |
 |  | Chains were run for 30,000 iterations and 10% was discarded as burn-in. |  |
CPGLMM ‡ | cplm | Compound Poisson generalized linear mixed model fit using maximum | [44] |
 |  | likelihood. |  |
DESeq2* | DESeq2 | DESeq2 analysis using default settings. To account for correlation, | [15] |
 |  | subjects are included as fixed effects in the model by replacing the |  |
 |  | random effects in Equation 4 with fixed effects. |  |
DESeq2 | DESeq2 | DESeq2 analysis using default settings, including only fixed effects in | [15] |
 |  | Equation 4. Correlation is ignored. |  |
edgeR* | edgeR | edgeR analysis using default settings. To account for correlation, | [14] |
 |  | subjects are included as fixed effects in the model by replacing the |  |
 |  | random effects in Equation 4 with fixed effects. |  |
edgeR | edgeR | edgeR analysis using default settings, including only fixed effects in | [14] |
 |  | Equation 4. Correlation is ignored. |  |
limma | limma | Count data were transformed for analysis with limma using voom. | [26] |
 |  | The duplicateCorrelation function using subject id as the blocking | [27] |
 |  | factor was used to account for correlation. |  |
LMM | lmerTest | Data were first transformed using DESeq2’s variance stabilizing | [45] |
 |  | transformation. Linear mixed models, which assume a normally |  |
 |  | distributed errors, were then fit to the transformed counts. |  |
MACAU | MACAU2 | Poisson GLMM with random effects to account for over-dispersion and | [19] |
 |  | correlation among observations. A known covariance matrix must be supplied; |  |
 |  | we used the covariance of the centered and scaled gene expression matrix as |  |
 |  | shown in the third example of [19]. |  |
NBGLMM | glmmADMB | Frequentist negative binomial generalized linear mixed model fit using | [46] |
 |  | maximum likelihood. |  |
ShrinkBayes | ShrinkBayes | Bayesian hierarchical model using shrinkage priors to control false | [23] |
 |  | discovery rates. We used a negative binomial distribution for the count |  |
 |  | data and shrinkage priors for all model parameters involved in statistical |  |
 |  | tests. All other parameters were set to default values. |  |
VarSeq ‡ | tcgsaseq | Model free method for identifying gene sets whose expression changes | [25] |
 |  | over time. Counts are transformed to log(CPM) for analysis. The |  |
 |  | asymptotic test was used to generate p-values. |  |