I have a set of eight gene clusters which I obtained from clustering a network. However, I had to update the genes in the network at a later stage which lead to certain deletions (genes not present in the new gene list). I did the clustering again and this time there were a total of 9 clusters. Is there some way in which I could compare the earlier eight clusters with the new nine ones to effectively say that "clustering has happened along the same lines". It would have been easier to do an eight to eight comparison checking from the largest to the smallest cluster. However, since I have nine clusters now, I am kinda in a fix as how to compare. I used the same clustering algorithm to cluster the network in both the cases

  • $\begingroup$ Did you try to assess the stability of the number of clusters? I haven't done any, but they are designed to help deciding if the number of clusters is correct or not. $\endgroup$ – llrs Oct 25 '18 at 7:51
  • $\begingroup$ @Llopis I looked at the best resolution values while doing the clustering in GEPHI. and tried the different clustering algorithms in CYTOSCAPE and picked the one with the best score again.Is there a separate method to assess the stability of the clusters because checking resolution is what is usually done. $\endgroup$ – The Last Word Oct 25 '18 at 13:44
  • $\begingroup$ Perhaps I'm not understanding correctly what you are doing. But these stability test I am mentioning are used when the number of clusters is unknown, so you test the resolution, or other metrics for a range of numbers of clusters. Then you choose the best one (either the one that is better by the score, or the one that increasing the number of clusters don't improve your metric much more) $\endgroup$ – llrs Oct 25 '18 at 15:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.