Identification of transcription factors dictating blood cell development using a bidirectional transcription network-based computational framework [CAGEseq_MLLAF9-AML]

Genome-wide transcription factor binding profiles provide valuable insights into hematopoietic development and malignancies. Today, advanced computational methods exploit cis-regulatory information to predict complex gene regulatory networks controlled by transcription factors. However, these prediction methods still require large amounts of reference data or experimental expertise. Here, we present a low-requirement network-based computational framework that exploits bidirectional transcription marked regulatory elements to identify important transcription factors driving hematopoietic decision making. Using CAGE-seq we predicted TF binding and confirmed experimentally validated TF important in cell conversion strategies by exploiting bidirectional regions. Next, we applied our framework to predict driving transcription factors in induced normal erythropoiesis and early myelopoiesis, as well as in acute myeloid leukemia associated MLL-AF9-driven immortalisation. Our approach allowed the identification of experimentally validated as well as thus far unexplored transcription factors in these processes. Overall design: Comparative CAGE transcription start sites (CTSS) expression profiling analysis of CAGE-seq data for primary CD14+ monocytes (FANTOM5 database) and MLL-AF9 AML.