See edgeR user manual for examples with the plotMDS function: That would roughly do a multidimensional scaling based on the top varying genes and indicate whether one of your sample depart from the others. As well as how the samples group together.

The first suggestion is similar to @massyah's, which is to say that there are things you can do to see how similar your data sets are to each other. A naive clustering of your data, for example, would hopefully but the replicates from the same condition next to each other.

In addition to the edgeR::plotMDS suggestion, you can look through the vignette for DESeq2. Section 8 of that vignette has some suggestions you can follow to explore the quality of your data.

Also, you might consider including a covariate to indicate the batch of the experiment when doing your linear modeling/differential expression testing to account for such effects.

For a quick and dirty try you can upload your count data to Scotty and we'll cluster it for you and you can see if you have an outlier replicate. Sometimes test and controls don't group together though. It depends on how much of an effect there is.

http://euler.bc.edu/marthlab/scotty/scotty.php

Did all of the library preps use exactly the same protocols? I only ask because things can change a lot in 9 months.

edgeR, DESeq and DEGSeq all can use counts as input. You can do a pairwise comparison between the replicates and get some idea.