6.0 years ago by
Washington University, St Louis, USA
How many normal samples do you want? You will find it very difficult to find any breast cancer (or any other cancer) expression datasets with substantial numbers of normal tissue. These are (unfortunately) rarely collected in cancer expression studies. Part of the reason is that doctors do not want to expose patients to unnecessary procedures. Another problem is that it is rarely clear what the correct matched normal tissue should be. Tumors are often a proliferation of certain types of cells which have different proportions of cell types compared to surrounding normal tissue. Also, there are microenvironment effects resulting from normal cells living in proximity to the tumor which can effect their transcript expression. Basically it is just really difficult to know what "normal" really is. For these reasons (and cost) it is common for normals to be excluded from RNA isolation plans. Very frustrating.
Having said all that. Searching through Oncomine I found 12 breast cancer studies (out of X) which report having some kind of normal comparator. Of those, only 3 have >100 samples. None of them are listed as Affy in Oncomine but I didn't have a chance to dig up their platforms. Maybe across the 12 studies you will be able to find a common platform for enough to combine and make some decent numbers. Long shot I would guess.