Skip to main content
Fig. 4 | BMC Medical Genomics

Fig. 4

From: A systematic analysis of genomics-based modeling approaches for prediction of drug response to cytotoxic chemotherapies

Fig. 4

Feature selection methods and controls: a-b Spearman Correlation Coefficients for different feature selection methods NLSVR (a), PCR (b). c-d Spearman Correlation Coefficients for control models NLSVR (c), PCR (d) the placement of the symbol indicates the mean with the ends representing the range. e Cluster Entropy (Sc), indicative of how well cell lines of the same histotype cluster using k-means. Comparable Sc as well as little difference in r indicate that histotype recognition drives model performance. Sc is relative to the random control (RCTR) where Sc = 1, perfect clustering would have a Sc=0. The asterisks indicate significance (p < 0.05) between the method and alternative models, indicated by color (black indicates is was significant compared to all other methods), using a non parametric Wilcoxon match-paired rank test. The calculated p values can be found in Additional file 3

Back to article page