Biology Reference
In-Depth Information
the following equation (gene symbols stand for their log expression and
coefficients are omitted for clarity):
Y
i
=
ESR1
+
ERBB2
+
AURKA
+
PLAU
+
STAT1,
(3)
where the response variable Y
i
is the expression of gene
i
. This model is
fitted separately for each gene
i
in the array. The association between
gene
i
and prototype
j
conditional on all other prototypes is tested using
the
t
-statistic for each coefficient. Because the
t
-statistics for different
datasets have different degrees of freedom, we put them all on the same
scale by transforming to the corresponding cumulative probabilities and
then to
z
-scores using the inverse standard normal cumulative distribu-
tion function.
The linear model in Eq. (3) is fitted separately to each gene in each
dataset, and the
z
-scores are combined meta-analytically using the inverse
normal method [Eq. (1)].
7
Genes with large values of
-
are assigned
to the module for which the association is highest. A stringent criterion
for |
-
| is used to maintain interpretability and to keep the modules to a
manageable size.
7.2.3. Coexpression patterns
Coexpression patterns of those genes assigned to modules are shown in
heat maps for two of the datasets (NKI and EMC) in Fig. 2, which also
shows survival information for each patient. There are three major
subgroupings of samples, corresponding to combinations of conven-
tional markers. The tumors (columns) were sorted first according to
breast cancer subtypes — (1) basal-like ER
−
/ERBB2
−
, (2) ERBB2
+
, and
(3) luminal ER
— and then, within each subtype, according
to the average expression of proliferation genes in the AURKA module.
For clarity, the five modules are horizontally separated in the figure.
Each module contains highly correlated or anticorrelated genes, as
shown by the vertical color patterns. The annotation of the modules
shows that they correspond well to the expected biological processes. The
strong banding patterns show that the modules corresponding to estrogen
+
/ERBB2
−
