Transcription clusters and developmental pathways - nature, nurture, noise

Establishing how the genomic DNA sequence determines cell fate is a grand challenge in biology. It is usually approached from the viewpoint that each gene is transcribed independently of others. However, there is increasing evidence that clusters of RNA polymerases (variously referred to as transcription factories, condensates and hubs) make most RNA. Here, I use this cluster-based view to present alternative approaches to the grand challenge of linking DNA sequence and cell fate. Artificial intelligence-based tools are driving stunning advances in predicting transcriptional outputs, which in turn direct cell fates; however, they are limited by the curses of dimensionality, data sparsity and understandability. I explore how these AI tools could be used to exploit under-appreciated but information-rich inputs provided by DNA:DNA contacts in clusters.