Correlation of gene signatures with in vitro antiproliferative data.

We performed unsupervised hierarchical clustering based on gene signatures for the 3 cancer panels. Significant (p<0.05) or marginally significant (p<0.1) p values are listed for the cell line panels where we identified clusters of cell lines with different sensitivities. For the EMT pathway we tested the predictive power of the expression profiles based on the elastic net regression model. The predicted and measured values (IC50) were correlated based on the Pearson correlation coefficient. In cases where significant correlations existed, p values are listed. Significance of EMT pathway clustering was confirmed for breast cancer by qPCR (p = 0.05). *confirmed by TLDA (NS indicates not significant p values>0.1). Correlation of gene signatures with in vitro antiproliferative data.