2-colour (channel) Agilent data

Question

[parinv]

I downloaded Agilent 2 color data of Agilent_sure-print_g3_ge_8x60k platform and performed differential Expression I followed pipeline given here. Now I want to further analyze it using WGCNA. After normalization by normalizeWithinArrays, the data is transformed to MAlist. There are Two matrix in the MA list. First is M matrix that contains log2 expression values and second is A matrix that contains average log2 expression values. My question is which values should be considered for further analysis?

The MA list object looks like :

    An object of class "MAList"
$targets
                                                                       FileName
Sample1
Sample2
Sample3

$genes
  Row Col ControlType       ProbeName  SystematicName
1   1   1           1 GE_BrightCorner GE_BrightCorner
2   1   2           1      DarkCorner      DarkCorner
3   1   3           1      DarkCorner      DarkCorner
4   1   4           0    A_23_P326296       NM_144987
5   1   5           0    A_24_P287941       NM_013290
62971 more rows ...

$source
[1] "agilent"

$printer
$ngrid.r
[1] 1

$ngrid.c
[1] 1

$nspot.r
[1] 384

$nspot.c
[1] 164


$M
                      Sample1                                     Sample2
[1,]                            0.6499983                             0.5357494
[2,]                            0.1989391                            -0.1607216
[3,]                           -0.1824825                            -0.6443034
[4,]                           -0.9675712                            -1.8801010
[5,]                            0.4388240                             1.0480026
 62971 more rows ...

$A
                    Sample1                                               Sample2
[1,]                            14.565923                             14.605850
[2,]                             2.897886                              3.441465
[3,]                             3.408228                              3.090076
[4,]                             6.348938                              6.201913
[5,]                             7.635765                              6.971435
    62971 more rows ...

Thank you

Parin

[Ram]

Hi,

Can you edit your post and format the code part with the 101010 button please? You'll need to copy the code and paste it again, as you used blockquote formatting instead of code formatting and that has ruined your original paste. Use the 101010 button, not the double quote button.

[parinv]

I updated the post, is it proper now?

[Ram]

Not yet. You'll need to select all the code content and hit the button, but I can do that for you now, as the blockquote format had mangled your content earlier.

[parinv]

Thank you, I will keep this in mind for any future question.

Answer 1

[Kevin Blighe]

Hi, in most / all use-cases, you will want the values held in M. The 2-colour arrays are obviously fundamentally different from the single-colour. The M values are the log2 ratios between the DNA of the sample under study and the control. Being acutely aware of your experimental design is important, though. Can you elaborate on it?

Be sure to remove those control probes, by the way, i.e., the ones with a value of 1 for ControlType.

Kevin

[parinv]

I am using this dataset for meta-analysis with other 3 datasets. Only this dataset was 2-colour array, while others are single-colour. I will filter the control probes.

Thanks a lot.

[parinv]

just one more doubt. while filtering the the probes in single-colour channel, we also filter probes that are abovebackground, what is the criteria to choose array number for this filter?

> IsExpr <- rowSums(y$other$gIsWellAboveBG > 0) >= 4

[Kevin Blighe]

You can choose any number. 4 was the number taken from the limma manual pages based on the fact that the study only had 8 samples in total (I think). So, if you have 20 samples, perhaps choose 5 or 10 (25% and 50% of samples, respectively.)

[parinv]

Thank you so much for your help.