Data downloaded from:

    http://www.broadinstitute.org/cgi-bin/cancer/publications/pub_paper.cgi?paper_id=43


GEDM.csv contains probes values from "Train dataset" data file.
LABELS.csv contains labels from "Samples table" data file.

Because data are not (e.g. rma) normalized (but used with original scanned
values), only the "Train dataset" is used, because the "Test dataset" is
composed by an independent and not omogeneous collection (see original paper).

Note that LABELS.csv contains information also for the "Test dataset" here
excluded.
