Hello,
I want to see how many members falls in one cluster when i set a specific distance distance to find out the similar elements in the matrix.
I did clustering using R for the following matrix and i want only one cluster when i set certain cutoff for the tree.
X1 X2 X3 X4 X5 X6 X7
1 0.000 0.170 6.256 6.252 6.245 6.224 6.300
2 0.170 0.000 6.249 6.245 6.237 6.217 6.292
3 6.256 6.249 0.000 0.232 0.309 0.328 0.277
4 6.252 6.245 0.232 0.000 0.280 0.274 0.269
5 6.245 6.237 0.309 0.280 0.000 0.257 0.254
6 6.224 6.217 0.328 0.274 0.257 0.000 0.246
7 6.300 6.292 0.277 0.269 0.254 0.246 0.000
And when i set the cut off 3 or any other number, i get small clusters formed
x <- rect.hclust(hr, h=3)
The cluster forms are
[[1]] X1 X2 X15 1 2 15
[[2]] X3 X4 X5 X6 X7 X8 X10 X12 X13 X14 3 4 5 6 7 8 10 12 13 14
[[3]] X9 X11 9 11
Now should i choose the cluster with highest number of members in the cluster? i.e. x[[2]]? Because i want only one cluster when i set certain cutoff for the tree.
Any help would be appreciated! Thanku
0 answers
No answers yet.
Log in to answer this question.
The question isn't easily answerable at the moment -- if you tell us what sort of data do you have, why are you clustering it and what you hope to do with the data in your cluster someone might be able to help
Clustering is only based on certain cutoffs like given a tree length what is the cluster?
You are cutting the hierarchical tree at a certain level to produce a specified number of clusters. And from these clusters, you want to choose only one cluster? Like David said, we don't know how to choose the cluster if we don't know what or why you are choosing the cluster for.
Okay... let me edit my question then
But why do you only want one cluster when you set a cut-off? All the clusters show similar elements in their own respective groups. What is this one cluster supposed to show?
what i want is when i set certain threshold how many elements are falling into the threshold? So, could i say that all the elements though it is in different cluster falls into the threshold?
What biological system or question are you trying to answer? We are really in the dark here as to what you want to achieve.
these are the RMSD value between the structures, i want to cluster the ones which are similar based on certain RMSD cutoff.
Grosy - have you plotted you clustering dendrogram? Doing that with
rect.hclustshould be clear what the clustering and the cut are doing, which might help you decide one what to do next.