Hello List,

I have a question about 'reciprocity' using maSigPro.
My dataset includes three time points (shared starting time 0h, 24h and 43h) with 5 different conditions (4 infection scenarios and one control). 
I first tried with a design matrix as below:

         Time Replicates Control ME49_1.2 ME49_3 RH_1.2 RH_3
U0_0a       0          1       1        1      1      1    1
U0_0b       0          1       1        1      1      1    1
U0_0c       0          1       1        1      1      1    1
U24_0a     24          2       1        0      0      0    0
U24_0b     24          2       1        0      0      0    0
U24_0c     24          2       1        0      0      0    0
U43_0a     43          3       1        0      0      0    0
U43_0b     43          3       1        0      0      0    0
U43_0c     43          3       1        0      0      0    0
M24_1.2a   24          4       0        1      0      0    0
M24_1.2b   24          4       0        1      0      0    0
M24_1.2c   24          4       0        1      0      0    0
M24_3a     24          5       0        0      1      0    0
M24_3b     24          5       0        0      1      0    0
M24_3c     24          5       0        0      1      0    0
M43_1.2a   43          6       0        1      0      0    0
M43_1.2b   43          6       0        1      0      0    0
M43_1.2c   43          6       0        1      0      0    0
M43_3a     43          7       0        0      1      0    0
M43_3b     43          7       0        0      1      0    0
M43_3c     43          7       0        0      1      0    0
R24_1.2a   24          8       0        0      0      1    0
R24_1.2b   24          8       0        0      0      1    0
R24_1.2c   24          8       0        0      0      1    0
R24_3a     24          9       0        0      0      0    1
R24_3b     24          9       0        0      0      0    1
R24_3c     24          9       0        0      0      0    1
R43_1.2a   43         10       0        0      0      1    0
R43_1.2b   43         10       0        0      0      1    0
R43_1.2c   43         10       0        0      0      1    0
R43_3a     43         11       0        0      0      0    1
R43_3b     43         11       0        0      0      0    1
R43_3c     43         11       0        0      0      0    1

And performed the analysis, using get.siggenes with vars=groups, to obtain a number of significantly-differentially-expressed genes for each comparison to the Control, e.g. sigs$sig.genes$ME49_1.2vsControl 

I then tried the analysis again, using a different design matrix, using a different sample (ME49_1.2) as the "reference":
 

         Time Replicates ME49_1.2 ME49_3 RH_1.2 RH_3 Control
U0_0a       0          1        1      1      1    1       1
U0_0b       0          1        1      1      1    1       1
U0_0c       0          1        1      1      1    1       1
U24_0a     24          2        0      0      0    0       1
U24_0b     24          2        0      0      0    0       1
U24_0c     24          2        0      0      0    0       1
U43_0a     43          3        0      0      0    0       1
U43_0b     43          3        0      0      0    0       1
U43_0c     43          3        0      0      0    0       1
M24_1.2a   24          4        1      0      0    0       0
M24_1.2b   24          4        1      0      0    0       0
M24_1.2c   24          4        1      0      0    0       0
M24_3a     24          5        0      1      0    0       0
M24_3b     24          5        0      1      0    0       0
M24_3c     24          5        0      1      0    0       0
M43_1.2a   43          6        1      0      0    0       0
M43_1.2b   43          6        1      0      0    0       0
M43_1.2c   43          6        1      0      0    0       0
M43_3a     43          7        0      1      0    0       0
M43_3b     43          7        0      1      0    0       0
M43_3c     43          7        0      1      0    0       0
R24_1.2a   24          8        0      0      1    0       0
R24_1.2b   24          8        0      0      1    0       0
R24_1.2c   24          8        0      0      1    0       0
R24_3a     24          9        0      0      0    1       0
R24_3b     24          9        0      0      0    1       0
R24_3c     24          9        0      0      0    1       0
R43_1.2a   43         10        0      0      1    0       0
R43_1.2b   43         10        0      0      1    0       0
R43_1.2c   43         10        0      0      1    0       0
R43_3a     43         11        0      0      0    1       0
R43_3b     43         11        0      0      0    1       0
R43_3c     43         11        0      0      0    1       0

Here, I found that the genes identified as significant in this group were different (i.e. sigs2$sig.genes$ControlvsME49_1.2)

In other words, sigs2$sig.genes$ControlvsME49_1.2 is not the same as sigs$sig.genes$ME49_1.2vsControl

even though the data is the same, and the comparison should be the same. 

Can you help?

Many thanks.