Machine Learning Uncovers the Transcriptional Regulatory Network for the Production HostStreptomyces albidoflavus

Abstract

AbstractStreptomyces albidoflavusis a popular and genetically tractable platform strain used for natural product discovery and production via the expression of heterologous biosynthetic gene clusters (BGCs). However, its transcriptional regulatory network (TRN) and its impact on secondary metabolism is poorly understood. Here we characterized its TRN by applying an independent component analysis to a compendium of 218 high quality RNA-seq transcriptomes from both in-house and public sources spanning 88 unique growth conditions. We obtained 78 independently modulated sets of genes (iModulons) that quantitatively describe the TRN and its activity state across diverse conditions. Through analyses of condition-dependent TRN activity states, we (i) describe how the TRN adapts to different growth conditions, (ii) conduct a cross-species iModulon comparison, uncovering shared features and unique characteristics of the TRN across lineages, (iii) detail the transcriptional activation of several endogenous BGCs, including surugamide, minimycin and paulomycin, and (iv) infer potential functions of 40% of the uncharacterized genes in theS. albidoflavusgenome. Our findings provide a comprehensive and quantitative understanding of the TRN ofS. albidoflavus, providing a knowledge base for further exploration and experimental validation.Graphical Abstract