The `MArrayLM2`

object returned by `dream()`

contains `cov.coefficients`

and `cov.coefficients.list`

as its components. `cov.coefficients`

is a single matrix, and `cov.coefficients.list`

is a list of matrices that are as many as the genes included in the analysis. The `?dream`

documentation has no explanation for these components, and the `?MArrayLM`

documentation of `limma`

states that `cov.coefficients`

is a "numeric matrix giving the unscaled covariance matrix of the estimable coefficients." That still doesn't make a clear sense to me because the covariance matrix of coefficients should be different for each gene, but `cov.coefficients`

is just a single matrix. I also checked if `cov.coefficients`

is the same as `cov(coefficients)`

(where `coefficients`

is another component that `lmFit()`

and `dream()`

return), but it was not the case. Can someone clarify what they really are?