Entering edit mode
Hi,
while I try to analysis my data as the following ,I faced some problem with (heatmap()):
> dat<-ReadAffy()
> dat
AffyBatch object
size of arrays=1164x1164 features (20 kb)
cdf=HG-U133_Plus_2 (54675 affyids)
number of samples=10
number of genes=54675
annotation=hgu133plus2
notes=
> dat2<-rma(dat)
Background correcting
Normalizing
Calculating Expression
> dat.m<-exprs(dat2)
The normalized data can be so large that clustering all the genes (or
arrays) becomes impossible. Clustering about 23000 genes takes about 1
GB of memory, and clustering 54675 genes would consume about more than 4 GBs of
memory, and would not be feasible on a standard Windows workstation.
arrays) becomes impossible. Clustering about 23000 genes takes about 1
GB of memory, and clustering 54675 genes would consume about more than 4 GBs of
memory, and would not be feasible on a standard Windows workstation.
So I tried to sample the data, and this sample
is then clustered. This should convey approximately the same information as
the clustering of the whole dataset:
is then clustered. This should convey approximately the same information as
the clustering of the whole dataset:
> n<-1:nrow(dat.m)
> n.s<-sample(n, nrow(dat.m)*0.1)
> dat.sample<-dat.m[n.s,]
> library(amap)
> clust.genes<-hcluster(x=dat.sample, method="pearson",
+ link="average")
> clust.arrays<-hcluster(x=t(dat.sample), method="pearson",
+ link="average")
> n.s<-sample(n, nrow(dat.m)*0.1)
> dat.sample<-dat.m[n.s,]
> library(amap)
> clust.genes<-hcluster(x=dat.sample, method="pearson",
+ link="average")
> clust.arrays<-hcluster(x=t(dat.sample), method="pearson",
+ link="average")
The sample size is here 10% of the original dataset.
Ok, then I tried to visualizing the clustering results as a heatmap:
> heatcol<-colorRampPalette(c("Green", "Red"))(32)
> heatmap(x=as.matrix(dat.m), Rowv=as.dendrogram(clust.genes),
+ Colv=as.dendrogram(clust.arrays), col=heatcol)
+ Colv=as.dendrogram(clust.arrays), col=heatcol)
Error in .heatmap(x=as.matrix(dat.m), Rowv=as.dendrogram(clust.genes),: row dendrogram ordering gave index of wrong length
<font face="monospace">Was that sample of the data make an error with heatmap()??</font>
<font face="monospace">Cheers,</font>
<font face="monospace">Tahani.</font>
Thank you dear.