Acute myeloid leukemia (AML) is an aggressive type of leukemia, characterized by the accumulation of highly proliferative blasts with a disrupted myeloid differentiation program. Current treatments are ineffective for most patients, partly due to the genetic heterogeneity of AML. This is driven by genetically distinct leukemia stem cells, resulting in relapse even after most of the tumor cells are destroyed. Thus, personalized treatment approaches addressing cellular heterogeneity are urgently required. Reconstruction of Transcriptional regulatory Networks (RTN) is a tool for inferring transcriptional activity in patients with various diseases. In this study, we applied this method to transcriptome profiles of AML patients to test if it provided additional information for the interpretation of transcriptome data. We showed that when RTN results were added to RNA-seq results, superior clusters were formed, which were more homogenous and allowed the better separation of patients with low and high survival rates. We concluded that the external knowledge used for RTN analysis improved the ability of unsupervised machine learning to find meaningful patterns in the data.
Keywords: AML; RNA-seq data; RTN; acute myeloid leukemia; reconstruction of transcriptional regulatory networks; transcription factors.