Integrative clustering of multi-level ‘omic data based on non-negative matrix factorization algorithm
Fig 2
Comparison of intNMF and iCluster over varying k.
First row represents the cluster prediction index, second row represents the plot of proportion of deviance (POD) given by iCluster method and third row represents adjusted rand index between (i) true and intNMF-clusters (red), (ii) true and iCluster-clusters (blue) and (iii) intNMF-clusters and iCluster-clusters (green). The POD is expected to result in minimum at true number of clusters. In other plots, maximum is expected at true number of clusters.