Dual threshold optimization and network inference reveal convergent evidence from TF binding locations and TF perturbation responses

Kang, Yiming; Patel, Nikhil; Shively, Christian A.; Recio, Pamela Samantha; Chen, Xuhua; Wranik, Bernd J.; Kim, Griffin; McIsaac, R. Scott; Mitra, Robi D.; Brent, Michael R.

doi:10.1101/gr.259655.119

Cited by 25 publications

(60 citation statements)

References 70 publications

(92 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…S8). Thus, the strength and location of the binding signal are meaningful predictors of whether significantly bound genes will respond to the perturbation, consistent with our earlier findings (Kang et al 2020). [-200, -100] bin.…”

Section: In Yeast Cells Tf Binding Locations and Strengths Discriminsupporting

confidence: 90%

“…Data on where in the genome each TF binds were expected to be of great value in determining its targets, but multiple studies have shown that, in the available large ChIP-chip and ChIP-seq datasets, the genes in whose regulatory DNA a TF binds do not correspond well to those that respond to perturbation of the TF (Gitter et al 2009;Lenstra and Holstege 2012;Cusanovich et al 2014;Kang et al 2020). We followed up on these observations by training machine learning models to predict which genes would respond to perturbation of a TF, given data on the TF's binding locations and several features reflecting the gene's epigenetic context.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Predicting which genes will respond to perturbations of a TF: TF-independent properties of genes are major determinants of their responsiveness

Kang

Brent

2020

Preprint

Self Cite

View full text Add to dashboard Cite

Background: The ability to predict which genes will respond to perturbation of a TF's activity serves as a benchmark for our systems-level understanding of transcriptional regulatory networks. In previous work, machine learning models have been trained to predict static gene expression levels in a given sample by using data from the same or similar conditions, including data on TF binding locations, histone marks, or DNA sequence. We report on a different challenge -- training machine learning models that can predict which genes will respond to perturbation of a TF without using any data from the perturbed cells. Results: Existing TF location data (ChIP-Seq) from human K562 cells have no detectable utility for predicting which genes will respond to perturbation of the TF, but data obtained by newer methods in yeast cells are useful. TF-independent features of genes, including their preperturbation expression level and expression variation, are very useful for predicting responses to TF perturbations. This shows that some genes are poised to respond to TF perturbations and others are resistant, shedding significant light on why it has been so difficult to predict responses from binding locations. Certain histone marks (HMs), including H3K4me1 and H3K4me3, have some predictive power, especially when downstream of the transcription start site. In human, the predictive power of HMs is much less than that of gene expression level and variation. Code is available at https://github.com/yiming-kang/TFPertRespExplainer. Conclusions: Sequence-based or epigenetic properties of genes strongly influence their tendency to respond to direct TF perturbations, partially explaining the oft-noted difficulty of predicting responsiveness from TF binding location data. These molecular features are largely reflected in and summarized by the gene's expression level and expression variation.

show abstract

Section: In Yeast Cells Tf Binding Locations and Strengths Discriminsupporting

confidence: 90%

Section: Discussionmentioning

confidence: 99%

Predicting which genes will respond to perturbations of a TF: TF-independent properties of genes are major determinants of their responsiveness

Kang

Brent

2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…We can broadly categorize timecourses at a dataset level based on existing knowledge. While strong acute regulation events are frequently associated with the direct binding of the induced TF (Kang et al, 2020), over 75% of genes responding in our dataset are new regulatory connections (Appendix Fig S10). Additionally, we find that 79% of genes reported as being directly bound by a TF based on published ChIP measurements do not exhibit a significant expression response in the corresponding TF's induction experiment (Appendix Fig S10; Teixeira et al, 2018).…”

Section: Transcriptional Responses Vary In Amplitude Kinetics and Smentioning

confidence: 99%

“…The final measurement (the gene expression profile) is the asymptotic readout of many layers of regulation and unobserved (but potentially relevant) molecular interactions. Using methods like ChIP-seq or ChIP-exo provides another strategy for determining GRNs that focus on transcription factors (TFs) and their target genes (Harbison et al, 2004;Gerstein et al, 2010;Nègre et al, 2011;Kheradpour & Kellis, 2014;Kim et al, 2014;Kang et al, 2020). Target genes with similar ChIP profiles can exhibit opposite expression responses (Lickwar et al, 2012), and highly expressed portions of the genome can exhibit strong ChIP signal even among unrelated proteins (Teytelman et al, 2013).…”

Section: Introductionmentioning

confidence: 99%

Learning causal networks using inducible transcription factors and transcriptome‐wide time series

Hackett

Baltz

Coram

et al. 2020

Molecular Systems Biology

Self Cite

View full text Add to dashboard Cite

We present IDEA (the Induction Dynamics gene Expression Atlas), a dataset constructed by independently inducing hundreds of transcription factors (TFs) and measuring timecourses of the resulting gene expression responses in budding yeast. Each experiment captures a regulatory cascade connecting a single induced regulator to the genes it causally regulates. We discuss the regulatory cascade of a single TF, Aft1, in detail; however, IDEA contains > 200 TF induction experiments with 20 million individual observations and 100,000 signal‐containing dynamic responses. As an application of IDEA, we integrate all timecourses into a whole‐cell transcriptional model, which is used to predict and validate multiple new and underappreciated transcriptional regulators. We also find that the magnitudes of coefficients in this model are predictive of genetic interaction profile similarities. In addition to being a resource for exploring regulatory connectivity between TFs and their target genes, our modeling approach shows that combining rapid perturbations of individual genes with genome‐scale time‐series measurements is an effective strategy for elucidating gene regulatory networks.

show abstract

“…We anticipate improvements coming from better network maps. One likely source of better maps is new, more accurate methods for measuring TF binding locations (53,(75)(76)(77)(78). The input network could also be improved by obtaining TF perturbation data from cells grown in new conditions.…”

Section: Discussionmentioning

confidence: 99%

Inferring TF activities and activity regulators from gene expression data with constraints from TF perturbation data

Brent²

2020

Preprint

Self Cite

View full text Add to dashboard Cite

Background: The activity of a transcription factor (TF) in a sample of cells is the extent to which it is exerting its regulatory potential. Many methods of inferring TF activity from gene expression data have been described, but due to the lack of appropriate large-scale datasets, systematic and objective validation has not been possible until now. Results: Using a new dataset, we systematically evaluate and optimize the approach to TF activity inference in which a gene expression matrix is factored into a conditionindependent matrix of control strengths and a condition-dependent matrix of TF activity levels. These approaches require a TF network map, which specifies the target genes of each TF, as input. We evaluate different approaches to building the network map and deriving constraints on the matrices. We find that such constraints are essential for good performance. Constraints can be obtained from expression data in which the activities of individual TFs have been perturbed, and we find that such data are both necessary and sufficient for obtaining good performance. Remaining uncertainty about whether a TF activates or represses a target is a major source of error. To a considerable extent, control strengths inferred using expression data from one growth condition carry over to other conditions. As a result, the control strength matrices derived here can be used for other applications. Finally, we apply these methods to gain insight into the upstream factors that regulate the activities of four yeast TFs: Gcr2, Gln3, Gcn4, and Msn2. Evaluation code and data available at https://github.com/BrentLab/TFA-evaluation Conclusions: When a high-quality network map, constraints, and perturbation-response data are available, inferring TF activity levels by factoring gene expression matrices is effective. Furthermore, it provides insight into regulators of TF activity.

show abstract

Dual threshold optimization and network inference reveal convergent evidence from TF binding locations and TF perturbation responses

Abstract: Service Email Alerting click here. top right corner of the article or Receive free email alerts when new articles cite this article-sign up in the box at the http://genome.cshlp.org/subscriptions

Cited by 25 publications

References 70 publications

Predicting which genes will respond to perturbations of a TF: TF-independent properties of genes are major determinants of their responsiveness

Predicting which genes will respond to perturbations of a TF: TF-independent properties of genes are major determinants of their responsiveness

Learning causal networks using inducible transcription factors and transcriptome‐wide time series

Inferring TF activities and activity regulators from gene expression data with constraints from TF perturbation data

Contact Info

Product

Resources

About