Skip to main content

Table 7 Summary of clustering results with dataset III

From: Microarray data mining: A novel optimization-based approach to uncover biologically coherent structures

Cluster Group Cluster Size -log10(P) Values Correction Correlation Precision
   Max. Min. Ave. Max. Min. Ave. Ave. Ave.
  B 61 4 2.5 4.2 1.2 1.7 0.609 0.753
  D 175 8 5.5 13.7 1.5 3.9 0.362 0.641
  F 102 102 2.9 2.9 2.9 0.9 0.172 0.494
A Initial 271 2 20.0 140.1 0a 19.7 0.655 0.461
  Final 271 2 21.7 140.1 1.8b 20.6 0.707 0.522
C   116 2 10.4 33.3 3.2 9.5 0.672 0.735
E Initialc - - - - - - - -
  Final 88 2 4.4 11.4 1.1 3.4 0.635 0.440
  1. Genes in dataset III are clustered by EP_GOS_Clust through a sequential process outlined in Figure 3. Genes in cluster groups A and E are further clustered by the iterative algorithm, yielding an initial and final set of clusters. Precision is defined as the fraction of genes within a cluster assigned to the predominant functional group within that cluster.
  2. aThe cluster p-value is zero if a GO search did not manage to uncover any significant annotation.
  3. bAfter iteratively clustering 184 genes into 15 initial clusters ('A' on Figure 3), just one poor cluster remains. The next worse cluster has a -log10(P) value of 4.1.
  4. cThere are no applicable initial values here since the remaining genes to be clustered are subjected to the second filter before being re-clustered into the initial 6 clusters (see Figure 3).