SEQanswers K-means clustering and heatmap
 User Name Remember Me? Password
 Register FAQ Members List Calendar Search Today's Posts Mark Forums Read

 Similar Threads Thread Thread Starter Forum Replies Last Post crazyhottommy Bioinformatics 6 11-24-2014 09:45 AM crazyhottommy Bioinformatics 3 08-20-2013 11:02 AM honey Bioinformatics 5 06-21-2012 02:12 PM yhong General 1 08-24-2011 09:10 AM Moodi Genomic Resequencing 2 01-24-2011 04:55 PM

 09-09-2013, 08:39 AM #1 crazyhottommy Senior Member   Location: Gainesville Join Date: Apr 2012 Posts: 140 K-means clustering and heatmap Hi everyone, I was plotting a ChIP-seq data using the pheatmap, see code below: km<- kmeans(m1,2) # determin how many cluster you want, I specify 2 here m.kmeans<- cbind(m1, km\$cluster) # combine the cluster with the matrix dim(m.kmeans) # [1] 903 602 # the last column is 602 o<- order(m.kmeans[,602]) # order the last column m.kmeans<- m.kmeans[o,] # order the matrix according to the order of the last column pheatmap( m.kmeans[,1:601], cluster_rows = F, cluster_cols = F, col= hmcols, breaks = bk, legend=FALSE, show_rownames=FALSE, show_colnames=FALSE) It works fine for me, I clustered the data to two groups by specifying K=2, the problem is that group 1 sometimes shows up in the upper part of the heatmap, sometimes it shows up in the bottom part of the figure if I plot it several times. I think it has to do with the assignment of the group number, say, the first group is assigned to 1, the other is assigned to 2. However, next time if you plot the same data, the first group assigned to 2, the other is assigned to 1. R randomly assigns the number to the groups. How can control this? Thank you every much!
 09-09-2013, 09:56 AM #2 JackieBadger Senior Member   Location: Halifax, Nova Scotia Join Date: Mar 2009 Posts: 380 I have had the same issues using k-means clustering and discriminate analysis of principle components (DAPC) package. While I can't answer your question, the package has a help mailing list that may be able to provide a solution.
 09-10-2013, 04:13 AM #3 jmw86069 Member   Location: RTP, NC, USA Join Date: Jun 2009 Posts: 28 Easiest fix is to call set.seed(123); before you run k-means, and/or before you run pheatmap. Depends upon which step(s) you want to be reproducible. If it is just the order on the heatmap itself, where you're not re-running k-means, you can also precompute the row dendrogram and send it to pheatmap. Doing it that way may be better for you in the long run since it gives you more control over how the clustering is scored and linked. I recommend hcluster() from the amap package, it is much faster than default hclust, as fast as other competing hclust replacements, but also offers scoring by a few types of correlation.
 02-04-2014, 09:12 PM #4 scyxr Junior Member   Location: iowa Join Date: Sep 2012 Posts: 2 The same issue for me all the time. How can we assign each cluster number such as cluster 1 ,cluster 2 into the heat map. Hopefully someones knowing about this could give us the help. Thx.