A sequence level model of an intact locus predicts the location and function of nonadditive enhancers

Abstract

Metazoan gene expression is controlled through the action of long stretches of noncoding DNA that contain enhancers—shorter sequences responsible for controlling a single aspect of a gene’s expression pattern. Models built on thermodynamics have shown how enhancers interpret protein concentration in order to determine specific levels of gene expression, but the emergent regulatory logic of a complete regulatory locus shows qualitative and quantitative differences from isolated enhancers. Such differences may arise from steric competition limiting the quantity of DNA that can simultaneously influence the transcription machinery. We incorporated this competition into a mechanistic model of gene regulation, generated efficient algorithms for this computation, and applied it to the regulation of Drosophila even-skipped (eve). This model finds the location of enhancers and identifies which factors control the boundaries of eve expression. This model predicts a new enhancer that, when assayed in vivo, drives expression in a non-eve pattern. Incorporation of chromatin accessibility eliminates this inconsistency.

Developmental Transcriptional Enhancers: A Subtle Interplay between Accessibility and Activity

“…[ 11,12 ] The fact that accessibility of enhancers is a strong predictor of their spatiotemporal activity has been also demonstrated with computational models. [ 10,13 ]…”

“…By determining boundaries of enhancer accessibility, PTFs might in fact define the exact composition of TFBS contributing to enhancer regulation. [ 13 ] Interestingly, in the context of D. melanogaster embryogenesis, the same factor that primes the direct decompaction of hundreds of enhancers, Zelda, also appears to seed higher‐order structures of the genome. [ 81 ] This is perhaps where the coordination of different levels of chromatin accessibility happens.…”

Fully Interpretable Deep Learning Model of Transcriptional Control

“…Coactivators turn a nearby quencher into an activator. This action is described by the equation [23,4]…”

“…Last, we sum the fractional occupancies of all activators remaining after the previous two steps. We consider the activators to lower the energy barrier in a diffusion limited Arrhenius rate law [4], which has the mathematical form of a sigmoidal thresholding function. This results an fully connected layer which yields the mRNA synthesis rate.…”

“…In this paper, we show that DNNs can be used to generate a predictive model of gene expression. Our starting point is a previously published model of transcriptional control [23], which is one of a family of so-called thermodynamic transcription models [38,21,42,40,22,16,23,32,41,4,5]. In these models, occupancy of DNA by TFs is calculated using thermodynamics, and phenomenological rules are used to calculate the transcription rate from the configuration of bound factors.…”