All three of these clustering methods demonstrated that promoter methylation signatures could accurately segregate the samples into the AML and ALL categories.
Unsupervised two-way hierarchical cluster analysis of both the entire set of the filtered array probes (35433 probes) and a subset of probes (371 probes) that showed the most overall variability (overall standard deviation >1.0) did not segregate the samples according to ascites volume (data not shown).
For each gene subset, we performed average-linkage hierarchical clustering which segregated the samples into two major groups based on the first bifurcation of the dendrogram.
Analysis of the resulting data by unsupervised hierarchical clustering segregated the samples into subpopulations defined by CD24, CD29 and H2BGFP expression.
PCA on all thirteen sequenced conditions segregates the samples based on growth factor treatment along PC1 (44.57% of the variance) and on inhibitor treatment along PC2 (17.72% of the variance).
As shown in Figure 2, 11 microRNAs were differentially expressed (false discovery rate (FDR) < 0.01) between these groups, and the expression values of these microRNAs in lung tissue segregated the samples from fibrotic and control mice.
Unsupervised clustering using the top 200 most-variable probes segregated the samples into primary breast tumors and bone metastases, irrespective of ER, PR, and Her2 status (data not shown).
Hierarchical clustering of our 84 tumor samples (excluding the samples from cluster-5), using the 17 common genes out of the 43 genes predictor, segregated the samples into two clusters, the first was composed mainly of tumors of the Low-stroma-subtype and the second was composed of tumors of the other subtypes.
Since Eschrich's and other published predictors mainly segregated the samples of the Low-stroma-subtype, next step was to address whether our Low-stroma-subtype predictor was able to identify in Eschrich's data set, the patients with good prognosis.
Next, we analysed this subgroup of genes in more detail, and this enabled us to segregate the tumour samples into three broad clusters, where all five genes are highly expressed (cluster a), all five genes show low expression (cluster b) and where there is variable expression of these genes (cluster c).
This amended ES expression set (designated the ES set without proliferation genes) could also segregate the 101 pretreatment tumor samples according to time to progression.
