Dear list,
I apologise in advance for being a limma amateur, but I hope you would help me with the following issue.
I am currently trying to analyse a reverse-phase protein array using limma. I am working with cell samples corresponding to 3 genotypes, and within each genotype there are several independent clones as well as technical replicates of some clones. The samples were collected over 3 days. I want to determine significant differences between the genotypes, but I realise that it is more statistically correct to tell the model that some clones are technical replicates + take into account the collection dates. I tried to do this with an appropriate design matrix. However, once I run lmFit, I get the highlighted error in the code below. I am confused as to why the two samples indicated would be problematic. I checked and there are no missing values for these particular samples in the original dataset.
Thank you in advance,
Ralitsa Madsen
> group.clone.date
[,1] [,2] [,3]
[1,] "WTC11_WT" "G7" "12dec"
[2,] "WTC11_WT" "B" "12dec"
[3,] "WTC11_WT" "G7" "14dec"
[4,] "WTC11_WT" "B" "15dec"
[5,] "WTC11_WT" "B" "15dec"
[6,] "WTC11_WT" "C" "15dec"
[7,] "WTC11_WT" "C" "15dec"
[8,] "WTC11_WT" "H11" "15dec"
[9,] "WTC11_WT" "H11" "15dec"
[10,] "WTC11_WT" "G7" "15dec"
[11,] "WTC11_H1047Rhet" "Het2" "12dec"
[12,] "WTC11_H1047Rhet" "Het3" "14dec"
[13,] "WTC11_H1047Rhet" "Het1" "14dec"
[14,] "WTC11_H1047Rhet" "Het3" "15dec"
[15,] "WTC11_H1047Rhet" "Het3" "15dec"
[16,] "WTC11_H1047Rhom" "G" "12dec"
[17,] "WTC11_H1047Rhom" "G" "14dec"
[18,] "WTC11_H1047Rhom" "F" "14dec"
[19,] "WTC11_H1047Rhom" "F" "15dec"
[20,] "WTC11_H1047Rhom" "F" "15dec"
[21,] "WTC11_H1047Rhom" "F" "15dec"
[22,] "WTC11_H1047Rhom" "G" "15dec"
>
> design.WTC11 <-
+ model.matrix( ~ 0 + group.WTC11 + clone + col.date)
> colnames(design.WTC11) <-
+ gsub("group.WTC11", "", colnames(design.WTC11))
> contr.matrix.WTC11 <- makeContrasts(
+ HOMvsWT = WTC11_H1047Rhom - WTC11_WT,
+ HETvsWT = WTC11_H1047Rhet - WTC11_WT,
+ HOMvsHET = WTC11_H1047Rhom - WTC11_H1047Rhet,
+ levels = colnames(design.WTC11)
+ )
> RPPA.fit <-
+ lmFit(mat_RPPA.limma.log, design.WTC11) # warning about partial NA coefficients
Coefficients not estimable: cloneG cloneHet3
Warning message:
Partial NA coefficients for 57 probe(s)
> sessionInfo()
R version 3.4.4 (2018-03-15)
Platform: x86_64-apple-darwin15.6.0 (64-bit)
Running under: macOS High Sierra 10.13.5
Matrix products: default
BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/3.4/Resources/lib/libRlapack.dylib
locale:
[1] en_GB.UTF-8/en_GB.UTF-8/en_GB.UTF-8/C/en_GB.UTF-8/en_GB.UTF-8
attached base packages:
[1] grid stats graphics grDevices utils datasets methods base
other attached packages:
[1] bindrcpp_0.2.2 stringr_1.3.1 ggthemes_3.5.0 cowplot_0.9.2 extrafont_0.17 gridExtra_2.3 lattice_0.20-35 ggplot2_2.2.1
[9] magrittr_1.5 dplyr_0.7.6 tidyr_0.8.1 RColorBrewer_1.1-2 gplots_3.0.1 edgeR_3.20.9 limma_3.34.9
(~3 years after...) Hi Aaron, I arrived to your answer by looking for help with a related question of mine and I'm wondering if you have any suggestions? I'm aware that I can just re-run this by considering
treatment
andtime
as covariates without having each factor independently specified, but perhaps there is a more direct way from the design? Perhaps you have any input to my question? Thanks a lot in advance