The ABA and Gensat expression data has been substantially compared and it does not match up all that well

An article by Zaldivar and Krichmar discusses a comparison and alignment of data (never an easy thing) between the Allen Mouse Brain Atlas and Gensat data, mainly. This article deals with the issue of data alignment and a real look at the two resources, gathering data from different sources with ABA a high throughput technique and Gensat was largely used for the myriad of ISH studies they had gathered.

The two sources are eerily different, see table 2 below.

Note, green means alignment, blue and red seem to dominate though!

My question when looking at something like this is how much overlap should we expect?
If there is not much overlap, then what conclusions should we make based on any one gene expression study?

By the way, the authors used the Neuroscience Information Framework to search gensat, but you can do your own comparison because NIF has an "integrated view" giving a quick overview of the data from Gensat, MGI and ABA.