Changes in version 1.23.1                        

    o   Sometimes prefit.beta was NaN, now properly handled in apeglm().
	Resolves some "Error in optimHess" issue propagated to DESeq2.

                        Changes in version 1.8.0                        

    o   Bringing in Josh Zitovsky's work on fast beta binomial
	GLM fitting in C++.

                        Changes in version 1.5.4                        

    o   Added possibility for weights and standard errors to bbEstDisp().
	The standard errors are for the log of dispersion (theta).

                        Changes in version 1.3.2                        

    o   Fix simple bug when no sample has only intercept coefficient.

                        Changes in version 1.2.0                        

    o   Faster fitting of Negative Binomial using RcppNumerical.

                        Changes in version 1.0.0                        

    o   apeglm is released on Bioconductor! Methods can be accessed
	from within DESeq2 by calling 'lfcShrink' with type="apeglm".