Table 4

Comparison of our methods and Haslett et al. [16]. Identification of differentially-expressed probesets validated by RT-PCR: given a chosen number of probesets selected by GFC, and the number of probesets validated by RT-PCR within this set, how many of the RT-PCR probesets were selected by the methods in this paper.
Method
 # of probesets selected:
# of GFC selections validated by RT-PCR
 Math
 Math
 Math
 Math
 Math
 Math
 Math
MAS5
 30a : 8b
 5c
 1
 5
 5
 5
 5
 4

50 : 11
 8
 2
 7
 7
 8
 7
 6

100 : 12
 11
 3
 10
 10
 11
 10
 11

139 : 13
 12
 4
 11
 11
 12
 12
 12
RMA
 30 : N/A
 2d
 1
 4
 2
 2
 2
 6

50 : N/A
 6
 2
 7
 6
 6
 6
 7

100 : N/A
 9
 4
 9
 9
 9
 8
 8

139 : N/A
 10
 4
 11
 10
 10
 10
 11

a The top 30 probesets were selected by each method.

b 8 probesets in the top 30 were validated by RT-PCR in Haslett et al. [16].

c 5 RT-PCR probesets were among the top 30 probesets selected by Math. The 5 RT-PCR probesets are among the 8 RT-PCR probesets selected by GFC.

d 2 RT-PCR probesets were among the top 30 probesets selected by Math (data normalized by RMA). The 2 RT-PCR probesets are among the 8 RT-PCR probesets selected by GFC, using the MAS5 normalization.

e Sensitivity parameter ν = 0.05.

Hu et al. BMC Genomics 2006 7:33   doi:10.1186/1471-2164-7-33