Fig. 1

Outline of the study design. Differentially expressed genes were identified with differential expression analysis. MiRWalk was used to reconstruct the functional miRNA-mRNA regulatory module. RSF method was then used with GMI signature to develop prognosis model in training cohort. The developed prognosis model was evaluated on the independent dataset. The gained RSF-based score was applied for survival analysis and patient stratification. Specifically, we removed 14 genes which were not found in 187 RNA-seq expression dataset and 100 miRNAs which were not found in 301 miRNA expression dataset. Samples from TCGA with no expression of the signatures genes/miRNAs were filtered out. One hundred forty-seven AML samples were selected based on the present expression values of 23 gene expression signatures from 187 RNA-seq expression dataset and 16 miRNAs from 301 miRNA expression dataset