Skip to main content
Fig. 2 | BMC Bioinformatics

Fig. 2

From: Meta-analysis of cell- specific transcriptomic data using fuzzy c-means clustering discovers versatile viral responsive genes

Fig. 2

Optimization of FCM parameters. a Average membership value (y-axis) per cluster with increasing fuzziness (x-axis), b Average number of genes per cluster (y-axis) for increasing fuzziness (x-axis) and four cluster association criteria, c 50 trials conducted with random initial assignment of the centroids found only 16% reproducible clusters, d Objective function values for FCM clustering with initial centroid assignment performed randomly  and by Ward’s method (red line) under fuzziness 1.1, 1.2 and 1.3 respectively. Ward based initialization converged more rapidly and produced stable and robust clustering solution

Back to article page