Arguments
counts
the counts of reads mapped to the gene. input as a G X S matrix, where G is the number of genes, and S is the number of samples
size
the normalization factors for the counts.
It should be a vector with length S, for example, the total number of reads for each column.
The default is Geometric Median of the counts in each column.
Users can also input the 'size' as a G X S matrix, so
group
a vector indicating the design of a 2-treatment assignment, for example group=c(1,1,2,2).
model
specify the discrete probability that model the counts.
We allow 'nbinom' and 'poisson' in our test, where 'nbinom' is the default choice that use negative-binomal model.
dispersion
the dispersion parameter for each gene (each row of the counts).
users can specify the estimates by their own method,
or by default, we will use quasi-likelihood method to estimate a dispersion for each gene