How to get MD5 sums on windows
certUtil -hashfile nameoffastq.gz MD5
or by instgalling md5sums.
Geoff Barton has an excellent dataset which he and collaborating groups used to test which (popular) tools to test for RNASeq DE analysis. (Arxiv) I shall say I told you so.
There is another paper coming to the same conclusion (cuffdiff2, deseq and edgeR only)
http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0103207
"Considering the overall performance based on three independent datasets in this study, Cuffdiff2 is not recommended for differential expression analysis at gene-level resolution, particularly if sequencing depth is low (i.e. <10 million reads per individual sample)."
The title says it all Sequencing technology does not eliminate biological variability
Should probably be repeated many times, that being said edgeR does a good job to actually do with really realy bad replication. (but it can't do magic no replication is no replication and technical doesn't count).