clustermap seems to reorder column labels even though col_cluster = False

I frequently use clustermap for visualisations of my data and couldn't do without it. However, I recently came across an oddity where the implementation seems to change the order of the column labels even though col_cluster = False. I suspect this has to do with the special structure of my column names (e.g. E14.5_Ash1l+/-_F). In detail, I try to visualise a precomputed z-score matrix clustered only by the rows to make differences between columns more visible and to identify row groups while keeping the columns as they are as they already have a predefined relationship to each other. However, the returned plot seems to have reordered the columns.

A quick check by manual plotting the same data using plt.imshow shows that only the labels seem to have changed ordering but not the data (compare the colors of the above to the below image)

So it seems that there is some part of the code that tries to sort the labels in some way even though they should stay in the given order. In turn this could result in wrong interpretations of the data.

I am using version seaborn 0.13.2.

mwaskom / seaborn

clustermap seems to reorder column labels even though col_cluster = False #3763