Fig. 2

Identification of top MRs in CC. a Left, Top 100 activated MRs and 100 suppressed MRs were selected within each of the five discovery datasets. The common 18 activated and 3 suppressed MRs were identified and shown as red and blue in the NES boxplot. Right, the same 18 and 3 MRs were shown in the NES boxplot of other 5 validation datasets. The NES value distribution of the 18 and 3 MRs compared to the rest of MRs were tested using Wilcoxon Rank-Sum test. The P-values are 2e− 13, 2e− 13, 2e− 13, 2e− 13, 2e− 13 for the 18 activated MRs for discovery set; 1e− 03, 1e− 03, 1e− 03, 1e− 03, 1e− 03 for the 3 suppressed MRs for the discovery set; 1e− 13, 4e− 13, 1e− 12, 3e− 12, 9e− 11 for the 18 activated MRs for validation set; 2e− 03, 1e− 03, 1e− 03, 2e− 03, 1e− 03 for the 3 suppressed MRs for the validation set. b The DA (NES scores) and DE (t-scores) matrix of the 21 MRs (18 activated and 3 suppressed) in the 5 discovery datasets. DA-DE Consistency set was selected using |DE| value> 3 and same sign as DA in all five datasets. MRs in signal pathway or transcription factors were highlighted in orange and green background respectively