Heatmap using pheatmap issues

Question

[1769mkc]

I have already asked this question how can I annotate my heat-map with respective genes or sample

mat<-matrix(ncol=2,nrow=4)
mat[,1]<-c(0.0387765,0.0306198,0,0)
mat[,2]<-c(0.0407321,0,0.00212375,0.0000484229)
colnames(mat)<-c("Value1","Value2")
rownames(mat)<-c("DDX11L1","AL627309.2","RP11-34P13.9","OR4F29")
pheatmap(mat)

this was the code i got and it works pretty good but since I gave that data was very small subset of my cuffdiff output .Now i want to do the same after filtering the cuffdiff output with some threshold value the dimension would be really huge.So I cannot use this code to construct the dataframe manually I suppose.I tried to convert the dataframe into matrix then add the gene name into the matrix when i try to plot I get this error

Error in hclust(d, method = method) :

I guess there is fundamentally wrong the way Im trying to do, any help or suggestion would be appreciated .

My filtered data filtered data set which I want to make heatmap

[Michael]

Your output is lacking the error message that normally follows the ':'

[1769mkc]

I tried with the normal heatmap function im getting the error "Error in heatmap(df) : 'x' must be a numeric matrix ". I guess as one of my column is character which contains the gene name ..

[Michael]

Likely, you should assign them using rownames and then remove that column, e.g.:

rownames(df) <- df[,1]
mat <- as.matrix(df[,-1])

[1769mkc]

rownames(df) <- df[,1] so you are assigning rownames to the data frame while selecting only the column?

[Michael]

Assuming, the row names are in the first column, yes. If in doubt, look at the output of head(df). When converting to matrix, all remaining columns must be numeric, as a matrix can only consist of a single element type, otherwise you get a character matrix.

[1769mkc]

thank you every much it worked....