Dear All,
I am working with proteome data which has quite a few missing values and I cannot afford to lose any proteins by excluding them. Additionally, I would not want to use any sort of imputation as it disrupts my data since it has been corrected for batch effects.
I would like to find the optimum number of clusters present in my data for which I am using ConsensusClusterPlus as it can deal with missing values. However, the PAC score keeps on decreasing with an increasing K, which I am aware can happen if the samples are not clearly clustering. I know M3C clustering is used popularly now-a-days, however, it cannot deal with missing values.I was wondering if I can get help to calculate these Reference Cluster stability Index (RCSI) used in M3C using the PAC scores that I have from consensus clustering?
I would be very happy to get some inputs on this or also if there are any better consensus clustering methods which can deal with missing values :)
Thank you very much!
Best,
Shweta