Figure 1
Bidirectional transcription initiation in cis-regulatory elements, including enhancers. (a) A schematic overview of the computational framework. First, bidirectional regions and gene expression are defined by CAGEfightR16. Gene expression is measured by summing all CAGE Tag Clusters (TCs) expressed per gene. Then, ANANSE is used to calculate a TF binding probability (binding score) by logistic regression using three types of data: enhancer intensity score in TPM (bidirectional regions), TF motif scores, and the average ReMap 2022 ChIP-seq coverage15,21, using Eq. (1). Gene expression (TPM), binding score, weighted distance to TSS, and genome-wide TF motif activity are then used to determine gene regulatory networks and its respective interaction score. The influence score is calculated based on differential expression (log2 fold change), edge distance from TF, and the interaction score of the source cell type and target cell type. The influence score represents how well the differences between two cell states can be explained by a TF, thus inferring importance for a cell’s state. The equations to calculate the scores, logistic regression, and GRNs were previously described by Xu et al.15. (b) Representation of summed CAGE-defined Transcription Start Sites (CTSS) expression at unidirectional (top) and bidirectional (bottom) transcription clusters. The orange color depicts the forward DNA strand ( + ) and in grey the reverse ( − ). (c) CAGE-seq tags representing bidirectional transcription in K562 cells for two separate genomic regions. (d) Bar graph representation of associated bidirectional regions with its closest gene. Y-axis shows the fraction of regions associated for its respective increment of kilobases (kb). (e) Number of bidirectional regions in K562 cells associated with various genomic locations. (f) Genome-wide co-occurrence (intersection size) of H3K4me1, H3K27ac, and bidirectional regions (Bi. regions) in K562 cells. The black dots represent intersection between the two ChIP-seq marks and bidirectional transcription. The set size represent the total amount of regions included.
