Table 1 

Summary of similarities between SAGE and Affymetrix HGU133GeneChips for the final dataset (= 1094) 

UP/DOWN classification 
Contingency table diagonal 
Pearson Correlation coefficient^{3} 
0–3 fold betweenratio difference 
N 



All transcripts 
63% 
76% 
0.453 
78% 
1094 
Low expresssion^{1} 
57% 
81% 
0.222 
78% 
572 
High expresssion^{1} 
69% 
81% 
0.578 
90% 
226 
Significant difference^{2} 
86% 
47% 
0.636 
70% 
167 


^{1. }Based on the binomial sampling error of SAGE tags, tags counts below 5.7 and 7.7 (per 100,000) for the WT and Stratagene sample, respectively, are not significantly different from tag count 0. When a tag falls below these thresholds in both libraries it is included in the "Low expression" group (line 2); when a tag counts is above these thresholds in both libraries it is included in the "High expression" group (line 3). The thresholds were calculated as the 95% confidence interval of the tag proportion: CI_{95%}) = n ± 1.96* with n = tag count; N = Library size and p = n/N (proportion) ^{2.}Significant difference between the two SAGE libraries is defined as a significant Pvalue (α<0.05) according to the Ztest between two libraries [21]. ^{3.}All observed correlation coefficients are significant at P < 0.01 

van Ruissen et al. BMC Genomics 2005 6:91 doi:10.1186/14712164691 