You are here:
Learning Boolean Networks in HepG2 cells using ToxCast High-Content Imaging Data (SOT annual meeting)
Citation:
Antonijevic, T. AND I. Shah. Learning Boolean Networks in HepG2 cells using ToxCast High-Content Imaging Data (SOT annual meeting). Presented at SOT 56th Annual Meeting and ToxExpo, Baltimore, MD, March 12 - 16, 2017. https://doi.org/10.23645/epacomptox.5178637
Impact/Purpose:
Poster presentation at the 2017 SOT annual meeting. Our findings illustrate the utility of BNs for learning mechanisms from time-course data that may differentiate between cellular programs involved in adaptation versus injury
Description:
Cells adapt to their environment via homeostatic processes that are regulated by complex molecular networks. Our objective was to learn key elements of these networks in HepG2 cells using ToxCast High-content imaging (HCI) measurements taken over three time points (1, 24, and 72h) and across 10 concentrations (0.39-200µM) for 309 chemicals. Cell states were monitored via phospho-p53 (p53), phospho-c-Jun (SK), phospho-Histone H2A.x (OS), phospho-Histone H3 (MA), phospho α-tubulin (Mt), mitochondrial membrane potential (MMP), mitochondrial mass (MM), cell cycle arrest (CCA), nuclear size (NS) and cell number (CN) endpoints. We used Boolean Networks (BNs) as a simple coarse-grained representation of biological regulatory networks. First, measured endpoints were standardized and then discretized into perturbed/unperturbed values based on a noise threshold (z0=1.28) and dynamic trends. Second, we inferred the best Boolean functions and constructed a set of 125 BNs for 2,193 trajectories with at least 1 perturbation using uniform sampling. The accuracy of initial 274,125 BNs was estimated as the number of errors between predicted and observed trajectories. 205,663 BNs with the smallest error, defined by the baseline error, were tested across analyzed trajectories. We defined “coverage” as the number of trajectories predicted by each BN with an accuracy ≤ to the baseline error. We found 610 BNs that covered all trajectories. The BN with the greatest coverage explained 1,446 trajectories, where: p53 and SK activation influences MA, NS, and CN, and OS in conjunction with NS regulate MM. These 1,446 trajectories were produced by low treatment concentrations that we believe represent cellular recovery processes. Trajectories produced by high concentration treatments that resulted in cell death were predicted by a diverse set of BNs. Our findings illustrate the utility of BNs for learning mechanisms from time-course data that may differentiate between cellular programs involved in adaptation versus injury.This abstract does not necessarily reflect U.S. EPA policy.
URLs/Downloads:
DOI: Learning Boolean Networks in HepG2 cells using ToxCast High-Content Imaging Data (SOT annual meeting)
ANTONIJEVIC - SHAH - ABSTRACT SOT 2017-1_FINAL.PDF (PDF, NA pp, 83.863 KB, about PDF)
TODOR-SOT2017-FINAL-1_M.PDF (PDF, NA pp, 1314.629 KB, about PDF)