Skip to main content
. 2019 Aug 6;14(8):e0220279. doi: 10.1371/journal.pone.0220279

Fig 3. Estimating the number of co-expression clusters.

Fig 3

Three methods were used to estimate the number of EdgeScaped clusters. (A) Calinski and Harabaz inter-cluster CH loss. CH loss value mapped between clusters for number of clusters ranging from 5 to 50 with the increment of 5. Loss value calculated for 500 thousand, 1 million, and 1.5 million sample data points. (B) Inter-cluster Silhouette Coefficient. Silhouette Coefficient mean value mapped between clusters for number of clusters ranging from 5 to 50 with the increment of 5. Loss value calculated for 500 thousand, 1 million, and 1.5 million sample data points. (C) Hierarchical clustering. Number of leaves obtained with 11 experimental iterations.