pheatmap with predefined row order (?)

[IsBeth]

Hello!

I used the pheatmap function (package pheatmap), an excellent function to obtain a heatmap of my top20 genes. But rows have to be shown within the heatmap in a specific order.

Now my ordered and rlog transformed dataframe looks something like this. It contains only 20 rows.

head(ordered_df)
control1 control2 treatment1 treatment2
gene1 -1.7645217 -1.7195209 -1.6657764 1.8478388
gene2 -1.8288594 -1.8174903 -1.8042068 1.8982069
gene3 -1.6516937 -1.6204703 -1.6078509 1.5970406
gene4 -1.3465805 -1.1734480 -1.4704845 1.2976865
gene5 -1.4837454 -1.0958759 -1.3840635 1.3615838
gene6 -1.0479161 -0.9808831 -1.0591158 0.9616953


I obtained the pheatmap with the following command:

pheatmap(ordered_df, annotation_col=annotcol)

I thought that it would work to simply preorder the data frame before giving it to the pheatmap function, but the heatmap shows rows in an incorrect order. Is there a way within the function to achieve this? It seems to me that there must be something very simple to do this, but I cannot find it.

[Zoltan]

Please check here the answer of D.R. on another forum: https://www.biostars.org/p/108247/

I tried to use
res <- pheatmap(matrix)
and than one may change value for
res$tree_row$order <- list.of.your.order
res$tree_col$order <- list.of.your.order or get the order from here,
but finally could not produce a heatmap from the modified res
D.R. has published a better solution