Melanie BODY - 2015-10-27

Good morning!

I am a new user of MeV and I need some help to figure out an issue that I haven't been able to fix by myself using MeV help documents, Google and forums...

I have a microarray dataset with 27 different treatments and 1276 genes. I ran 2 K-means clusterings using MeV with the following parameters:
Cluster: genes
Distance metric: Pearson correlation
Calculate: k-means
Cluster number: 5 and then 10
Iterations: 50

When I compared the number of genes in each cluster and the number on the centroid graphs, it didn't match. So, I looked at the gene list for all the clusters and found out that the same 98 genes are in all of the 5 (or 10) clusters!

For example:
Cluster 1: 290 genes on the centroid graph but 388 genes (290+98) in the cluster gene list
Cluster 2: 153 genes on the centroid graph but 251 genes (153+98) in the cluster gene list
Cluster 3: 271 genes on the centroid graph but 369 genes (271+98) in the cluster gene list
Cluster 4: 240 genes on the centroid graph but 338 genes (240+98) in the cluster gene list
Cluster 5: 224 genes on the centroid graph but 322 genes (224+98) in the cluster gene list
Total: 290+153+271+240+224 + 98 = 1276

I don't understand why those 98 genes appear in all of the 5 (or 10) clusters. Does someone have an idea?

Thank you in advance for your help!
Have a great day,

Melanie