I have missing samples in my interaction design, so I followed the tutorial to get rid of missing samples:

m1 = model.matrix(~Strain + Condition + Strain:Condition, meta)
all.zero <- apply(m1, 2, function(x) all(x==0))
idx <- which(all.zero)
m1 <- m1[,-idx]

But now I'm not sure how to supply this to create a DESeq object? Ive tried suppliying it to the design argument but get an error

dds <- DESeqDataSetFromMatrix(countData = raw_counts,
                              colData = meta, 
                              design = m1)

I can create the object using a different design, but then how do i replace the design with the correct model matrix? I might be confusing design and model.matrix, if so can somebody clarify?

my current solution is to use a ~1 design, and then supply the model.matrix to the full argument (even thought I am using the Wald test)

library("DESeq2")
dds <- DESeqDataSetFromMatrix(countData = raw_counts,
                              colData = meta, 
                              design = ~1)

m1 = model.matrix(~Strain + Condition + Strain:Condition, colData(dds))

all.zero <- apply(m1, 2, function(x) all(x==0))
all.zero
idx <- which(all.zero)
m1 <- m1[,-idx]


dds <- DESeq(dds, test = "Wald", betaPrior = FALSE, full = m1, parallel = TRUE)