A multi-scale transcriptional regulatory network knowledge base forEscherichia coli

Abstract

Uncovering the structure of the transcriptional regulatory network (TRN) that modulates gene expression in prokaryotes remains an important challenge. Transcriptomics data is plentiful, necessitating the development of scalable methods for converting this data into useful knowledge about the TRN. Previously, we published the PRECISE dataset for Escherichia coli K-12 MG1655, containing 278 RNA-seq datasets created using a standardized protocol. Here, we present PRECISE 2.0, which is nearly three times the size of the original PRECISE dataset and also created using a standardized protocol. We analyze PRECISE 2.0 at multiple scales, demonstrating multiple analytical strategies for extracting knowledge from this dataset. Specifically, we: (1) highlight patterns in gene expression across the dataset; (2) utilize independent component analysis to extract 218 independently modulated groups of genes (iModulons) that describe the TRN at the systems level; (3) demonstrate the utility of iModulons over traditional differential expression analysis; and (4) uncover 6 new potential regulons. Thus, PRECISE 2.0 is a large-scale, high-quality transcriptomics dataset which may be analyzed at multiple scales to yield important biological insights.

Mining all publicly available expression data to compute dynamic microbial transcriptional regulatory networks

“…One such function computes a clustered heatmap of iModulon activities to identify correlated groups of iModulons ( Figure S3 ). These iModulons often respond to a common stimulus, and represent a computational method to define stimulons 22,38 . Here, we guide readers through two stories that were discovered through iModulon visualizations.…”

“…Even though each iModulon contains a high fraction of genes regulated by SigB, none of the individual iModulons capture all of the genes thought to be regulated by SigB. Often, this separation represents independent regulatory modes for a single global regulator 22,23 . On the other hand, iModulons in the bottom right-quadrant provide evidence for new regulator binding sites.…”

“…iModulon activities have intuitive interpretations, leading to a dynamic representation of an organism’s transcriptional regulatory network (TRN) 10,11,19–21 . Even for pairwise comparisons, iModulons can simplify the analysis from thousands of differentially expressed genes to merely dozens of differentially activated iModulons 22,23 .…”

Machine Learning Uncovers a Data-Driven Transcriptional Regulatory Network for the Crenarchaeal Thermoacidophile Sulfolobus acidocaldarius

“…ICA decomposition of our RNA-seq compendium ( Figure 5A ) revealed 45 iModulons, which we ranked by explained variance in Figure 1D . iModulons with high explained variance are nearly always regulated by global transcriptional regulators ( Lamoureux et al, 2021 ). However, the iModulon that explained the largest variance in the S. acidocaldarius data was primarily enriched with uncharacterized or poorly characterized genes.…”

Machine learning uncovers a data-driven transcriptional regulatory network for the Crenarchaeal thermoacidophile Sulfolobus acidocaldarius

“…iModulons provide evidence of an uncharacterized global regulator ICA decomposition of our RNA-seq compendium revealed 45 iModulons, which we ranked by explained variance in Figure 1D. iModulons with high explained variance are nearly always regulated by global transcriptional regulators (Lamoureux et al, 2021). However, the iModulon which explained the largest variance in the S. acidocaldarius data was primarily enriched with uncharacterized or poorly characterized genes.…”