Mapping DNA sequence to transcription factor binding energy in vivo

Abstract

Despite the central importance of transcriptional regulation in biology, it has proven difficult to determine the regulatory mechanisms of individual genes, let alone entire gene networks. It is particularly difficult to decipher the biophysical mechanisms of transcriptional regulation in living cells and determine the energetic properties of binding sites for transcription factors and RNA polymerase. In this work, we present a strategy for dissecting transcriptional regulatory sequences using in vivo methods (massively parallel reporter assays) to formulate quantitative models that map a transcription factor binding site’s DNA sequence to transcription factor-DNA binding energy. We use these models to predict the binding energies of transcription factor binding sites to within 1 k B T of their measured values. We further explore how such a sequence-energy mapping relates to the mechanisms of trancriptional regulation in various promoter contexts. Specifically, we show that our models can be used to design specific induction responses, analyze the effects of amino acid mutations on DNA sequence preference, and determine how regulatory context affects a transcription factor’s sequence specificity.

The Energetics of Molecular Adaptation in Transcriptional Regulation

“…The disagreement is most evident for the weakest operator O3 [green lines in Fig. 4(A)], though we have discussed previously that the induction profiles for weak operators are difficult to This study accurately describe and can result in comparable disagreement for the wild-type repressor (10,22). Including D# AI as a perturbed parameter in addition to K A and K I improves the predicted profiles for all four mutants.…”

“…These binding energies are comparable to that of the wild-type repressor affinity to the native LacI operator sequence O3, with a DNA binding energy of 9.7 k B T. The mutation Q21M increases the strength of the DNA-repressor interaction relative to the wild-type repressor with a binding energy of 15.43 +0.07 0.06 k B T, comparable to the affinity of the wild-type repressor to the native O1 operator sequence ( 15.3k B T). It is notable that a single amino acid substitution of the repressor is capable of changing the strength of the DNA binding interaction well beyond that of many single base-pair mutations in the operator sequence (4,22).…”

Logomaker: Beautiful sequence logos in python

“…Logomaker thus fills a major need in the Python community for flexible logo-generating software. Indeed, preliminary versions of Logomaker have already been used to generate logos for multiple publications (Belliveau et al, 2018;Wong et al, 2018;Forcier et al, 2018;Nguyen et al, 2018;Barnes et al, 2019;Kinney and McCandlish, 2019). Logomaker is thoroughly tested, has minimal dependencies, and can be installed from PyPI by executing pip install logomaker at the command line.…”

Biophysical models of cis-regulation as interpretable neural networks

“…We can infer thermodynamic models like these for a cis-regulatory sequence of interest (the wild-type sequence) from the data produced by a massively parallel reporter assay (MPRA) performed on an appropriate sequence library [25]. Indeed, a number of MPRAs have been performed with this explicit purpose in mind [25,[27][28][29][30]. To this end, such MPRAs are generally performed using libraries that consist of sequence variants that differ from the wild-type sequence by a small number of single nucleotide polymorphisms (SNPs).…”

“…2B consistently took about 15 minutes on a standard laptop computer. The model fitting procedure in [25], by contrast, relied on a custom Parallel Tempering Monte Carlo algorithm that took about a week to run on a multi-node computer cluster (personal communication), and more recent efforts to train biophysical models on MPRA data have encountered similar computational bottlenecks [29,30].…”