heatmap
1
0
Entering edit mode
@kfbargadehues-1528
Last seen 9.6 years ago
Dear list, I have obtained a heatmap from my list of significant genes using the function heatmap(). I enclose an image of it. The sequences are coloured using maPalette(). I was confused when looking at the gene tree structure, some of the sequences that are coloured in red on the left hand side of the tree (hopefully because of their high intensity) cluster together with some of the sequences that are coloured in green on the same samples. Should I not see the "red-on- the-left-group/green- on-the-right-group" sequences all clustered together on one big branch of the genetree, and the "green-on-the-left-group/red-on-the-right- group" sequences on another big branch? In a different way, why would a sequence cluster tightly together with others that have a completely swapped colour profile across samples? Hope I make myself understandable Best, David
• 1.1k views
ADD COMMENT
0
Entering edit mode
@kfbargadehues-1528
Last seen 9.6 years ago
Thanks for your answer Kevin, I used the default option, which I think is euclidean (is this what you call absolute?) since heatmap uses dist() arguments. I checked ?dist and found that Spearman and Pearson (which are the ones that I am most familiar with) were not among the options for distance measures. Is there a reason for this? maybe they can be found elswhere? David > What are you using as the measure of similarity? If you use absolute > correlation, then things that are highly negatively correlated (since > they have opposite expression patterns) should cluster together. > Kevin > > kfbargad at ehu.es wrote: > > Dear list, > > > > I have obtained a heatmap from my list of significant genes using the > > function heatmap(). I enclose an image of it. The sequences are > > coloured using maPalette(). > > > > I was confused when looking at the gene tree structure, some of the > > sequences that are coloured in red on the left hand side of the tree > > (hopefully because of their high > > intensity) cluster together with some of the sequences that are > > coloured in green on the same samples. Should I not see the "red- on- > > the-left-group/green- > > on-the-right-group" sequences all clustered together on one big branch > > of the genetree, and the "green-on-the-left-group/red-on-the- right- > > group" sequences on another big branch? In a different way, why would > > a sequence cluster tightly together with others that have a completely > > swapped colour profile across samples? > > > > Hope I make myself understandable > > > > Best, > > > > David > > > > > > ------------------------------------------------------------------- ----- > > > > _______________________________________________ > > Bioconductor mailing list > > Bioconductor at stat.math.ethz.ch > > https://stat.ethz.ch/mailman/listinfo/bioconductor >
ADD COMMENT
0
Entering edit mode
Yes, there is a reason. In fact, if the arrays v and w are normalized (mean(v) = mean(w) = 0, sd(v) = sd(w) = 1), then the euclidean distance between w and v is, in fact, 2 - 2 cor(u, v). The closer cor(u,v) is to 1, the smaller the euclidean distance... Carlos J. Gil Bellosta http://www.datanalytics.com El lun, 06-02-2006 a las 17:43 +0100, kfbargad at ehu.es escribi?: > Thanks for your answer Kevin, > > I used the default option, which I think is euclidean (is this what > you call absolute?) since heatmap uses dist() arguments. I checked > ?dist and found that Spearman and Pearson (which are the ones that I > am most familiar with) were not among the options for distance > measures. Is there a reason for this? maybe they can be found elswhere? > > David > > > What are you using as the measure of similarity? If you use absolute > > correlation, then things that are highly negatively correlated > (since > > they have opposite expression patterns) should cluster together. > > Kevin > > > > kfbargad at ehu.es wrote: > > > Dear list, > > > > > > I have obtained a heatmap from my list of significant genes using > the > > > function heatmap(). I enclose an image of it. The sequences are > > > coloured using maPalette(). > > > > > > I was confused when looking at the gene tree structure, some of > the > > > sequences that are coloured in red on the left hand side of the > tree > > > (hopefully because of their high > > > intensity) cluster together with some of the sequences that are > > > coloured in green on the same samples. Should I not see the "red- > on- > > > the-left-group/green- > > > on-the-right-group" sequences all clustered together on one big > branch > > > of the genetree, and the "green-on-the-left-group/red-on-the- right- > > > group" sequences on another big branch? In a different way, why > would > > > a sequence cluster tightly together with others that have a > completely > > > swapped colour profile across samples? > > > > > > Hope I make myself understandable > > > > > > Best, > > > > > > David > > > > > > > > > ------------------------------------------------------------------- > ----- > > > > > > _______________________________________________ > > > Bioconductor mailing list > > > Bioconductor at stat.math.ethz.ch > > > https://stat.ethz.ch/mailman/listinfo/bioconductor > > > > _______________________________________________ > Bioconductor mailing list > Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor
ADD REPLY
0
Entering edit mode
Hi David, To have a more homogeneous genetree cluster I 'pimp' the heatmap function with the option available for it. My heatmap function that I use look like that: > heatmap(x, col=cols, distfun=function(c){Dist(c,method="pearson")}, labCol=colnames(cd.exprs), Colv=NA, Rowv=cov(cd.exprs), hclustfun=function(c){hclust(c,method='mcquitty')}) The distance function 'Dist' come from the package 'amap'. But the important parameters are the reordering paramater 'Rowv' read the help page of heatmap for more explanation, and the method use by the hierarchical clustering to order the leave between them, here 'mcquitty' (?hclust for more infos) I hope that help... At least for me it gave more homogeneous hierarchical cluster. David On Feb 6, 2006, at 17:43, kfbargad at ehu.es wrote: > Thanks for your answer Kevin, > > I used the default option, which I think is euclidean (is this what > you call absolute?) since heatmap uses dist() arguments. I checked > ?dist and found that Spearman and Pearson (which are the ones that I > am most familiar with) were not among the options for distance > measures. Is there a reason for this? maybe they can be found elswhere? > > David > >> What are you using as the measure of similarity? If you use absolute >> correlation, then things that are highly negatively correlated > (since >> they have opposite expression patterns) should cluster together. >> Kevin >> >> kfbargad at ehu.es wrote: >>> Dear list, >>> >>> I have obtained a heatmap from my list of significant genes using > the >>> function heatmap(). I enclose an image of it. The sequences are >>> coloured using maPalette(). >>> >>> I was confused when looking at the gene tree structure, some of > the >>> sequences that are coloured in red on the left hand side of the > tree >>> (hopefully because of their high >>> intensity) cluster together with some of the sequences that are >>> coloured in green on the same samples. Should I not see the "red- > on- >>> the-left-group/green- >>> on-the-right-group" sequences all clustered together on one big > branch >>> of the genetree, and the "green-on-the-left-group/red-on-the- right- >>> group" sequences on another big branch? In a different way, why > would >>> a sequence cluster tightly together with others that have a > completely >>> swapped colour profile across samples? >>> >>> Hope I make myself understandable >>> >>> Best, >>> >>> David >>> >>> >>> ------------------------------------------------------------------- > ----- >>> >>> _______________________________________________ >>> Bioconductor mailing list >>> Bioconductor at stat.math.ethz.ch >>> https://stat.ethz.ch/mailman/listinfo/bioconductor >> > > _______________________________________________ > Bioconductor mailing list > Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor >
ADD REPLY

Login before adding your answer.

Traffic: 906 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6