Is there any chance to solve "the model matrix is not full rank" in DESeq2?

My design matrix is a count matrix, one row per gene and one column per sample,

The sample information (colData in dds) is a dataframe with two column, a column for condition containing pre and post treatment and column containing subject Id.

I am asking this question with regards to do paired samples

The problem is happened when

dds_1 <- DESeqDataSetFromMatrix(countData = df, colData = sample_info, design = ~ subject + condition)

This is the error I get:

Error in checkFullRank(modelMatrix) : the model matrix is not full rank, so the model cannot be fit as specified. Levels or combinations of levels without any samples have resulted in column(s) of zeros in the model matrix.

I read the vignette section 'Model matrix not full rank': but I could not find any solution for the problem.

Is there any solution for this problem?

You need to update your post by giving an example of your design matrix

You have two pieces of information. The subject and the condition/time. As Basti already noted, condition and time are identical, and you can only use one (e.g., pre and before are the same thing and post and after are the same thing as well). What you are trying to do is determine peptides that are differentially abundant at the two time points, adjusting for subject-specific differences.

As an aside, if you have m/z counts, then using DESeq2 is OK, but if you actually have something like normalized area under an integrated peak, then you should instead use limma, most likely limma-trend. Analyzing proteomics data is non-trivial, and it's pretty easy to do something really suboptimal. Anyway...

You need a Sample info data.frame that looks more like this:

> head(Sample_info, 10)
   Subject Condition
1        1       pre
2        2       pre
3        1      post
4        2      post
5        3       pre
6        4       pre
7        3      post
8        4      post
9        5       pre
10       6       pre

And now this will create a full rank model matrix

> library(limma)
> design <- model.matrix(~Subject + Condition, Sample_info)
> is.fullrank(design)
[1] TRUE

And the final column of that design matrix will test for the difference between pre and post after adjusting for subject.