Abstract
MotivationTo understand the regulatory pathways underlying diseases, studies often investigate the differential gene expression between genetically or chemically differing cell populations. Differential expression analysis identifies global changes in transcription and enables the inference of functional roles of applied perturbations. This approach has transformed the discovery of genetic drivers of disease and possible therapies. However, differential expression analysis does not provide quantitative predictions of gene expression in untested conditions. We present a hybrid approach, termed Differential Expression in Python (DiffExPy), that uniquely combines discrete, differential expression analysis with in silico differential equation simulations to yield accurate, quantitative predictions of gene expression from time-series data.ResultsTo demonstrate the distinct insight provided by DiffExpy, we applied it to published, in vitro, time-series RNA-seq data from several genetic PI3K/PTEN variants of MCF10a cells stimulated with epidermal growth factor. DiffExPy proposed ensembles of several minimal differential equation systems for each differentially expressed gene. These systems provide quantitative models of expression for several previously uncharacterized genes and uncover new regulation by the PI3K/PTEN pathways. We validated model predictions on expression data from conditions that were not used for model training. Our discrete, differential expression analysis also identified SUZ12 and FOXA1 as possible regulators of specific groups of genes that exhibit late changes in expression. Our work reveals how DiffExPy generates quantitatively predictive models with testable, biological hypotheses from time-series expression data.Availability and implementationDiffExPy is available on GitHub (https://github.com/bagherilab/diffexpy).Supplementary information Supplementary data are available at Bioinformatics online.
Highlights
Aberrant regulation of gene expression is frequently associated with diseases; changes to gene expression serve as key proxies to infer cell state (Emilsson et al, 2008)
DiffExPy identifies (i) minimal dynamical systems models that accurately predict gene expression dynamics in untrained conditions, (ii) specific Gene Ontology (GO) terms associated with classes of expression dynamics and (iii) specific transcription factors (TFs) associated with genes with similar expression responses
We find that the mean absolute log-fold change (LFC) between the phosphoinositide 3-kinase (PI3K) KI and WT conditions correlates with improved model accuracy (Spearman’s q 1⁄4 0.636, P 1⁄4 5.4e-26, Supplementary Fig. S2)
Summary
Aberrant regulation of gene expression is frequently associated with diseases; changes to gene expression serve as key proxies to infer cell state (Emilsson et al, 2008). Differential gene expression analysis quantifies changes in gene expression between cell states. Expression is compared between genetically different cells, cells exposed to different exogenous treatments—such as small molecules, proteins, temperatures or other environmental cues—or a combination of several treatments. Each gene in the analysis is categorized as a differentially expressed gene (DEG) or not. This categorization is often based on the magnitude of the log-fold change (LFC) of its expression between experimental conditions and by an VC The Author(s) 2019.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
