Skip to main content

Table 2 Parameter filtering

From: Challenges in microarray class discovery: a comprehensive examination of normalization, gene selection and clustering

Removed method p-value Superior method
T1 15 9.468e-207 PC 5
T1 100 4.037e-127 PC 5
M 15 1.435e-111 PC 5
hclusteuclaverage 4.926e-100 kmeans
hclustmanhaverage 4.536e-97 kmeans
STD 15 1.556e-88 STD 1000
M 100 2.045e-76 M 1000
P0 9.134e-75 P1
PC 3 1.241e-50 PC 15
T1 1000 3.739e-36 PC 15
pameucl 4.036e-32 kmeans
PC 5 1.033e-21 PC 15
pamcorr 8.688e-18 hclustcorrward
som 1.340e-13 kmeans
hclustcorraverage 2.337e-10 hclustcorrward
M 1000 4.359e-07 PC 15
  1. In an iterative process pairwise comparisons of normalization, gene selection and clustering methods are made to remove less favorable methods. This table presents the removed methods, the uncorrected p-value resulting from a Wilcoxon one-sided test comparing the removed method with a superior method.