I'm playing around with DESeq2 and I see in the manual that the DESeq() function is a wrapper for estimateSizeFactors(), estimateDispersions() and nbinomWaldTest(). However, when I run the three commands separately I get slightly different results. Specifically the log-fold changes differ in the 2nd decimal place.

So, is DESeq() a wrapper for the functions or a different implementation of the functions? I would expect to get identical results regardless of which way I run DESeq()