Fig. 3
From: Large-scale literature mining to assess the relation between anti-cancer drugs and cancer types
Valication with IC-50 values. Left hand side: boxplot of the IC-50 values for cancer/compound combinations with low, high and very high association. For ’Overlap Publications’ and ’Overlap Genes’ grouping is based on \(-\text{log}_{10}\) Fisher test p-values (low: \(<1\), high: between 20 and 80 and very high \(\ge 80\)). For ‘Word Embedding’ grouping is based on cosine similarity (low: \(<0.1\), high: between 0.2 and 0.6 and very high \(\ge 0.6\)) Right hand side: boxplot of the IC-50 values for compound Nilotinib comparing leukemia derived cell lines with the remaining cell lines