[BioC] [marray] Importing SMD files into marray

[Jeff Gentry]

On Mon, 9 Jan 2006, Furge, Kyle wrote:
> If you are able to read one file at a time, but not as a collection, it may
> be that the arrays are from different print runs and the dimensions of the
> arrays are not be exactly the same.
> 
> The last time I remember looking at this, the .xls header contains the
> printrun identifier.  In addition, if you downloaded the experiment archive,
> it should contain information (i.e. .gal files) for each printrun used.

They don't seem to have the file (I've looked at this particular dataset
as well, it is the 2004 Zhao breast cancer paper, although she's having
similar problems with other sets from the SMD).

> Maybe look at the 18195 and 19938 arrays and determine if the layout is
> exactly the same.

I've not yet identified exactly what the difference (at least a meaningful
one) is between those files but removing 18195.xls seems to "solve" the
problem (in that the read.SMD() call finishes w/ a marrayRaw object, but I
don't know if that's an appropriate solution).  I saw a post on this list
from a year and a half ago from someone trying to read in this same
dataset (with read.maimages() from limma) and he identified the same file
as being the problematic one.  FWIW all but one of the other .xls files
list their Category as "Breast" and Subcategory as "primary tumor" while
18195 and one other file have "Method testing" and "Amplification and
Labeling".

-J