Skip to main content

Table 1 Comparisons of different methods used to identify IS5 neighbouring genes from microarray experiments. Results obtained when different test statistics (rank product – RP, median rank – MR, Mean, SAM and t-statistics) and normalization methods (global and scale normalization) were applied to experimental data from the control, isolate 'A' and isolate 'B'. These results were also compared to existing genome sequence data. Top 50 to 100 genes were predicted as IS5 neighbours from microarray experiments. The neighbours from sequence data were defined as 5 genes upstream and 5 genes downstream of the IS5 elements and IS5 itself, so there were 121 neighbours based on sequence data. The numbers in this table represent number of genes that overlap between the microarray experiment and the sequence data. The higher the number is, the better consistency between the microarray experiment and the sequence data.

From: Genome-wide localization of mobile elements: experimental, statistical and biological considerations

Method Top 50 Top 100
   control A B control A B
Rank – based RP 13 46 31 15 69 43
  MR 13 45 32 15 70 47
Global Normalization Mean 20 37 20 23 65 29
  SAM 20 38 24 23 65 29
  t 20 38 24 23 65 31
Scale Normalization Mean 20 44 21 23 65 23
  SAM 20 40 22 23 63 29
  t 20 16 24 23 30 31