Fwd: Re: Fwd: Re: [BioC] Heatmap function

Marcus marcusb at biotech.kth.se
Thu Oct 30 09:31:21 MET 2003

>Hello again... I still have some questions. If you have the time Sean (or 
>someone else) I would be very grateful.
>Yes, the M-values are log, but I have checked and I do not have any 
>infinite or NaN values in the matrix.
>I have som questions regarding hclust. It says that the function 
>calculates the distances row-wise in the helpfile. But what I do not 
>understand is that I have chips with M-values that are seperated 
>column-wise. Am I supposed to have them by row instead?
>Another question.
>The function looks like this hclust(as.dist(myobject),  method ="complete")
>The as.dist have euclidean distance by default and the hclust have 
>"complete" as default. What does that mean, that It calculates two?? 
>distances or that on distancemeasure is "shut off"?
>And the final question.
>If I use hclust to create an object to use in the heatmap function, is 
>that the object that I give to the heatmapfunction both as colv and rowv?
>heatmap(x=mymatrixwithcorrelations, Rowv = 
>Colv=distanceofmymatrixwithcorrelations, distfun=dist, hclustfun = hclust 
>... etc) ?
>Or is there a way to fist cluster by row in hclust and then by column?
>Thanks for the help
>/ A very confused person
>>>I assume that the M values are a log, so you might check to be sure that you
>>>do not have infinite or NaN results from taking logs of zero and the like.
>>>As for dist, use ?dist to get the help file for dist.  There are other
>>>distances that you can use (by changing the arguments to dist).  However,
>>>one trick that I have found useful is to do your clustering as you want it
>>>done before calling heatmap and then call heatmap with Rowv=rowclust and
>>>Colv=colclust where rowclust and colclust are "cluster" objects that you can
>>>obtain from calling hclust.  In this way, you can make your clusters "look
>>>good" before spending the time having heatmap do it for you each time.
>>>Hope that helps.  Oh, by the way, I am absolutely not an expert, so check
>>>with other folks, also.

Marcus Gry Björklund

Royal Institute of Technology
AlbaNova University Center
Stockholm Center for Physics, Astronomy and Biotechnology
Department of Molecular Biotechnology
106 91 Stockholm, Sweden

Phone (office): +46 8 553 783 39
Fax: + 46 8 553 784 81
Visiting adress: Roslagstullsbacken 21, Floor 3
Delivery adress: Roslagsvägen 30B

More information about the Bioconductor mailing list