get different heatmap from website and R

Question

[Rey]

Hi all,

I have differntial gene expression and I want to make a heat map. I put their expression on the clustvis website, and I got the different heat map from R. I have 142 genes and 4 samples. I used the R

this code:

####
file <- read.csv ("")
file <- file[order(),]
row.names() <- file$B
file <- [,2:5]
file_matrix <- data.matrix()
file_heatmap <- heatmap(_matrix, Rowv=NA, Colv=NA, col = cm.colors(256), sccale="column", margins=c(5,10))
###

the attached image I got (https://ibb.co/kyZ0af)

and the expression of gene 1 is

genes           S1                  S2              S3            S4
gene 1  174136.1093 160236.9196 145866.516  181905.6098
gene 18 52.56674663 79.04848213 55.4397647  1883.182956

the expression for gene18 in heatmap for S4 is white, but its expression in this sample is higher than others. the code and my heatmap are correct?

[sangram_keshari]

Your 4th column seems to be empty here, but the heatmap doesn't. Your code seems somewhat correct, but as its partially given can't say much.

Please describe correctly what you want to know?

[Rey]

I want to make a heatmap and cluster these four stages, and S4 is not empty. in heatmap gene 18 in the S4 is white, but as you could see its expression is higher than others. the similar pattern could be seen for gene 1, I want to know my heatmap is correct or not. As I am a beginner in R, I tried a lot to make a matrix for using pheatmap, but I couldn't so I used this easy code.

genes S1 S2 S3 S4 gene1 174136.1093 160236.9196 145866.516 181905.6098 gene18 52.56674663 79.04848213 55.4397647 1883.182956

[sangram_keshari]

Okay got your point. So basically you saying your 18th gene have a high degree of expression but in heatmap its showing white.

Let me tell you the heatmap you generated using the webtool is clustered one, So it may possible that the order of genes may vary. Thats why you seeing on 18th row a white scale which may belong to a different gene.

[Rey]

actually, this heatmap is generated by R, the heatmap based on clustvis website show different cluster and for gene 1 in S1 shows the lowest expression.

The image from the website: (https://ibb.co/fVRgY0) So I don't know which one is correct!! image from R or website

[Rey]

actually, this heatmap is generated by R, the heatmap based on clustvis website show different cluster and for gene 1 in S1 shows the lowest expression.

The image from the website: (https://ibb.co/fVRgY0) So I don't know which one is correct!! image from R or website

Answer 1

[cpad0112]

OP code does column wide scaling (incorrectly typed sccale in OP code). Try drawing heatmap with column wise scaled values (do not scale scaled values).

[Rey]

Is it possible to explain more detailed about "do not scale scaled values"?

[Kevin Blighe]

It makes little sense to scale values that are already scaled. Can you confirm that changing sccale to scale improves the situation?