Himanshu Ardawatia
2006-Apr-07 15:01 UTC
[R] Dealing with missing values in HeatMap generation
Hi,
I want to generate a heatmap for my data (in a matrix). However, the
data has some missing values (represented as blank).
I get the following errors (with the blanks and with blanks replaced by
NA and including the option rm.na = TURE):
> filename = "input_heatmap.txt"
> g <- as.matrix(filedata)
> fg <- rainbow(nrow(g), start=0, end=.3)
> gg <- rainbow(ncol(g), start=0, end=.3)
> hg <- heatmap(g, col = cm.colors(256), scale="column",na.rm =
TRUE,
+ RowSideColors = fg, ColSideColors = gg, margin=c(5,10),
+ xlab = "Average per Species", ylab= "Pathways",
+ main = "heatmap(<Ka/Ks Average Data per Species>, ..., scale =
\"column\")")
Error in heatmap(g, col = cm.colors(256), scale = "column", na.rm =
TRUE, :
'x' must be a numeric matrix
Is there anyway to deal with it?
Second question: What is the basis for generation of the dendrogram
(over the heatmap) in the heatmap? Is it simple hierarchical clustering?
Thanks in advance
Himanshu
Michael Dondrup
2006-Apr-07 15:58 UTC
[R] Dealing with missing values in HeatMap generation
Hi, 'g' must be a numeric matrix and obviously it's not (suspect it's character?). If 'g' is numeric in your example depends on how you read data into 'filedata' and this step is missing in your code (maybe you just forgot to use read.table?). See ?read.table for that, and use argument colClasses if required. Am Friday 07 April 2006 17:01 schrieb Himanshu Ardawatia:> Hi, > > I want to generate a heatmap for my data (in a matrix). However, the > data has some missing values (represented as blank). > I get the following errors (with the blanks and with blanks replaced by > > NA and including the option rm.na = TURE): > > filename = "input_heatmap.txt" > > g <- as.matrix(filedata) > > fg <- rainbow(nrow(g), start=0, end=.3) > > gg <- rainbow(ncol(g), start=0, end=.3) > > hg <- heatmap(g, col = cm.colors(256), scale="column",na.rm = TRUE, > > + RowSideColors = fg, ColSideColors = gg, margin=c(5,10), > + xlab = "Average per Species", ylab= "Pathways", > + main = "heatmap(<Ka/Ks Average Data per Species>, ..., scale > \"column\")") > Error in heatmap(g, col = cm.colors(256), scale = "column", na.rm = TRUE, > : 'x' must be a numeric matrix > > Is there anyway to deal with it? > > Second question: What is the basis for generation of the dendrogram > (over the heatmap) in the heatmap? Is it simple hierarchical clustering? > > Thanks in advance > Himanshu > > ______________________________________________ > R-help at stat.math.ethz.ch mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide! > http://www.R-project.org/posting-guide.html
On 4/7/06 11:54 AM, "Himanshu Ardawatia" <Himanshu.Ardawatia at bccs.uib.no> wrote:> Sean Davis wrote: > >> >> On 4/7/06 11:01 AM, "Himanshu Ardawatia" <Himanshu.Ardawatia at bccs.uib.no> >> wrote: >> >> >> >>> Hi, >>> >>> I want to generate a heatmap for my data (in a matrix). However, the >>> data has some missing values (represented as blank). >>> I get the following errors (with the blanks and with blanks replaced by >>> NA and including the option rm.na = TURE): >>> >>> >>>> filename = "input_heatmap.txt" >>>> g <- as.matrix(filedata) >>>> fg <- rainbow(nrow(g), start=0, end=.3) >>>> gg <- rainbow(ncol(g), start=0, end=.3) >>>> hg <- heatmap(g, col = cm.colors(256), scale="column",na.rm = TRUE, >>>> >>>> >>> + RowSideColors = fg, ColSideColors = gg, margin=c(5,10), >>> + xlab = "Average per Species", ylab= "Pathways", >>> + main = "heatmap(<Ka/Ks Average Data per Species>, ..., scale >>> \"column\")") >>> Error in heatmap(g, col = cm.colors(256), scale = "column", na.rm = TRUE, : >>> 'x' must be a numeric matrix >>> >>> >> >> We can't see from your example what "filedata" is, so we can't tell what "g" >> is. >> >> >> >>> Is there anyway to deal with it? >>> >>> >> >> You will need to assure that "g" is a numeric matrix. >> >> >> >>> Second question: What is the basis for generation of the dendrogram >>> (over the heatmap) in the heatmap? Is it simple hierarchical clustering? >>> >>> >> >> Reading the ?heatmap page will help here. It describes what the defaults >> for clustering actually are. >> >> Sean >> >> >> > Thanks for the quick reply! the data table is as follows (Here I have > put 'NA', but I could also have just a 'space' in other files): Please see. > > mouse mouse1 rat rat1 human human1 chimp > path1 0.12 0.32 0.48 1.2 0.67 0.58 0.01 > path2 0.13 0.90 0.09 1.6 0.23 0.50 0.09 > path3 0.20 0.32 0.05 0.45 0.89 0.34 0.04 > path4 0.23 0.65 0.92 0.67 0.78 0.78 0.91 > path5 0.89 0.01 0.01 1.01 'NA' 0.98 0.79 > path6 0.78 0.32 0.13 1.99 0.54 0.08 0.56 > path7 0.34 0.54 'NA' 0.32 0.70 0.45 0.34 > path8 0.15 0.86 0.85 0.56 0.97 0.86 0.11 > path9 0.10 0.12 0.08 2.01 0.37 0.57 0.04 >But from your code above, you never load these data. Sean