Fig. 2
From: Predicting molecular initiating events using chemical target annotations and gene expression
Example schematic of training data set structure for ESR-1/2 (−) classifier. Binary classifiers were trained using “MIE-Active” and “MIE-Inactive” collections of gene expression profiles that are equal in number. In this example, the MIE-Active set is comprised of profiles derived from chemical treatments that are associated with ESR1 or ESR2 inhibition based on RefChemDB annotations. The MIE-Inactive training set is comprised of an equal number of profiles derived from chemical treatments that have no annotation for ESR1 or ESR2 in RefChemDB