Time series comparison with DESEQ2

Question

[halo22]

Hello All,

For my analysis I have three time points Time1, Time2 and Time3 these were measured based upon three different treatments. I am trying to run differential analysis between T1 vs T2, T1 vs T3 and T2 vs T3, I have some technical replicates that I summed using the collapseReplicates() function, the "runsCollapsed" column below describes that.

For running the comparison I have tried using "~SampleName+Status" as my design matrix. This doesn't seem to work and I have the following error:

    Object <- DESeqDataSetFromMatrix(countData = CountFrame, colData = Pheno , design = 
   ~SampleName + Status)

  factor levels were dropped which had no samples
  Error in checkFullRank(modelMatrix) : 
  the model matrix is not full rank, so the model cannot be fit as specified.
  One or more variables or interaction terms in the design formula are linear
  combinations of the others and must be removed.


   Pheno file:
    Sample.ID   SampleName  Well    Status  runsCollapsed
    S11         MS1A         C4     Time1   C4,D4
    S13         MS1B         C5     Time2   C5,D5
    S1          MS2A         A1     Time1   A1,B1
    S3          MS2B         A2     Time2   A2,B2
    S21         MS5A         C9     Time1   C9,D9
    S23         MS5B         C10    Time2   C10,D10
    S25         MS5C         C11    Time3   C11,D11
    S33         MS7A         E4     Time1   E4,F4
    S35         MS7B         E5     Time2   E5,F5
    S37         MS7C         E6     Time3   E6,F6
    S39         MS8A         E7     Time1   E7,F7
    S41         MS8B         E8     Time2   E8,F8
    S43         MS8C         E9     Time3   E9,F9
    S45         MS9A         E10    Time1   E10,F10
    S47         MS9B         E11    Time2   E11,F11 
    S49         MS9C         E12    Time3   E12,F12

Can someone please suggest if there is a better way to do this? Also for Time3 I have lesser samples compared to Time1 and Time2. Could this be the reason for the error?

[swbarnes2]

I think you need to drop "Samplename" from your design, or edit the sample names so they don't have the letters at the end. You don't really have two dimensions of conditions here, because every sample name has one and only one status. If all your samples really are that different, and you don't want to combine them together, you have no replicates, which really limits what you can learn from this experiment.