Weighted gene
combinations separating each individual tumor from the
set of all normal tissues. We indicate the weight of each
gene along with it.
Clustering
trees for the various cancer types, along with weighted
combinations of genes at each cluster node.
In
the pictures representing the trees, the leaf nodes are
marked with the tumor ID, while the internal nodes are
marked <ID: Margin: Count>, where "ID" is the ID
of the internal node, Margin is the combined margin
of the sub-tree, and Count is the number of genes in the
weighted combination at that node. For a tree with k internal
nodes, the IDs of the internal nodes are integers in the
range 0..k-1.
The feature set shown in Table 3 is for
Breast cancer Cluster ID #21.
The lung cancer cluster tree is illustrated
in Figure 6 of the paper.
Expanded
feature sets using the quadratic programming method:
Breast Cancer Cluster ID #21: [PCL] [Heatmap]
These results are shown in Figure 9 in the paper.
Expanded gene-lists at the cluster nodes. The clusters are
indicated by the tumor IDs of the leaf nodes.
-
Breast
-
CNS
-
Kidney
-
Ovary
-
Pancreas
Comparison
of expansion methods shown in the Appendix:
Breast Cancer Cluster ID #1 with 3 tumors.
Feature Sets: [Maximum Margin]
[Quadratic
Programming] [Minimum Overlap
for 10 iterations]
