Pheatmap: change annotation relating to colData()

Question

I am trying to create a pheatmap from the pairwise comparisions of a DESEq object, but I want that the labels of rows and columns are related to the metadata, not to be sample names.

Experiment set-up: I simply have 3 different cell types with 3 replicates each Goal: make a heatmap where these cell types cluster according to the pairwise correlation values

What I have done so far and how my data looks like:

Data frame with sample information:

> sample_info[,1:3]
    Sample                Cell_type Treatment
AM1    AM1     Alveolar_macrophages   healthy
AM2    AM2     Alveolar_macrophages   healthy
AM3    AM3     Alveolar_macrophages   healthy
IM1    IM1 Interstitial_macrophages   healthy
IM2    IM2 Interstitial_macrophages   healthy
IM3    IM3 Interstitial_macrophages   healthy
T1      T1                  T_cells   healthy
T2      T2                  T_cells   healthy
T3      T3                  T_cells   healthy

Data frame with read counts:

> head(only_counts)
               AM1    AM2    AM3    IM1    IM2    IM3     T1     T2     T3
let-7a-1-3p   1383    930   1321    621    734    347    325    355    911
let-7a-2-3p      0      0      0      1      1      4      0      0      0
let-7a-5p   396731 293379 408655 177221 165887 152788 295030 331667 457912
let-7b-3p       40     24     23    151    172     87     81    140    170
let-7b-5p     5889   3051   2353  16342  15507  10990  31974  30384  34134
let-7c-1-3p      4      2      0    103     83     65      0      0      0

DESEq object:

dds <- DESeqDataSetFromMatrix(countData = only_counts,
                              colData = sample_info,
                              design= ~ Cell_type)

Transformations and calculations:

vsd <- varianceStabilizingTransformation(dds, blind=T)
vsd_mat <- assay(vsd) #extract the vst matrix
vsd_cor <- cor(vsd_mat) #compute pairwise correlation values

Heatmap

pheatmap(vsd_cor,
         main="Hierarchical clustering",
         annotation = colData(dds)$Cell_type)
Error in `[.default`(annotation_col, colnames(mat), , drop = F) : 
  incorrect number of dimensions

When I do the simple heatmap: pheatmap(vsd_cor, main="Hierarchical clustering")

But I want that instead of sample names, cell type is displayed (colData(dds)$Cell_type)

I have tried similar things but none worked:

pheatmap(vsd_cor,
             annotation = sample_info$Cell_type)

pheatmap(vsd_cor,
                  annotation_col = colData(dds)$Cell_type,
    annotation_row=colData(dds)$Cell_type)

pheatmap(vsd_cor,
                  annotation_col = sample_info$Cell_type,
    annotation_row=sample_info$Cell_type)

Answer 1

In case someone has the same question, I figured it out: After getting the correlation matrix:

rownames(vsd_cor) <- paste(sample_info$Cell_type)
colnames(vsd_cor) <- paste(sample_info$Cell_type)
pheatmap(vsd_cor,
         main="Hierarchical clustering")