Use Case

Mapping Transcriptome Data to Metabolic Models of Gut Microbiota

A. Douglas (Cornell) with J. Chaston (Brigham Young University), A. Moya (University of Valencia), G. Thomas (University of York), A. Heddi (INSA Lyon), and B. Barker (Cornell CAC)


  • Use host and gut-microbial express data to infer which metabolic reactions occur in each organism as a result of dietary inputs

Why Aristotle?

  • Heterogeneous instance types and sizes with high availability to allow for unpredictable computational demands
  • Allows generation of reproducible computational biology pipelines in the form of VMs or VM configurations


  • Created a Windows instance for MATLAB and a Samba server for file sharing across Windows and Linux.
  • Developed a transcriptomic pipeline for Docker.
  • Documented scalable file access for cross-cloud and many-instance file/app access.
  • Investigated Cornell Supercloud as possible means to scale out to other clouds.
  • Developed and published a computational model for the whiteflies system.
  • Analyzed and compared three independently-evolved communities in xylem-feeding insects. Generated advanced draft metabolic reconstructions for 5 bacteria needed to simulate Drosophila-microbial community metabolic interactions. The draft models of the 2 Lactobacilli have an average of 586 genes, 1193 metabolites and 750 reactions. The draft models of the 3 Acetobacter have an average of 686 genes, 1362 metabolites and 1129 reactions.
  • Initiated a computational strategy on Aristotle, based on a new optimization framework SteadyCom, to obtain the metabolic fluxes for optimized relative abundances of the partners under equilibrium conditions of constant growth rates.
  • Integrated the SteadyCom computational framework to model multi-species metabolic interaction in the insect gut.
  • Performed SteadyCom on test models in Aristotle using a newly confirmed Windows VM with MATLAB and the COBRA Toolbox.
  • Built a Linux VM with Docker, MATLAB, and Gurobi to containerize the simulation environment.
  • Perform a Flux Variability Analysis on the four microbes composing the Drosophila gut microbiome.
  • Constructed multi-species networks in silico of 2, 3, 4 and 5 bacterial species, quantified predicted metabolite fluxes, and wrote a MATLAB function to restore constraints to a multi-species model created from a single species.
  • Verified multi-species models behave as expected with standard flux algorithms.
  • Migrated a Windows VM to Linux VM in OpenStack, and wrote scripts and library functions to expedite running model simulations and choosing ideal model parameters.
  • Using SteadyCom, investigated the scope of metabolic interactions that occur among a Drosophila gut microbial community.
  • Identified 159 unique metabolites that are exchanged and showed that the gut microbial community is an important source of TCA cycle intermediates 2-oxoglutarate and succinate.
  • Aristotle VM enabled metabolic interaction research was published in mBio Journal.
  • Developed a new method, Semi Dynamic SteadyCom, to investigate priority effects among Drosophila gut microbes. Semi Dynamic SteadyCom employs SteadyCom with discrete time steps.
  • Developed a modeling framework with strong static typing, named hCOBRA, in the Haskell programming language. hCOBRA extends the COBRA Toolbox for MATLAB with the aim of making large scale simulations more reliable and more scalable.
  • Integrated functionality from the Funflow library into hCOBRA to allow caching of workflow results in a distributed setting.
  • Added R-based analyses to our hCOBRA workflows through the HaskellR library.
  • Expanded our analysis to investigate priority effects in a 5-member Drosophila gut microbe community comprising three Acetobacters and two Lactobacilli, which will require continued Aristotle cloud resources. Analyzed data from these simulations and prepared a manuscript.


Angela Douglas explains how exploiting animal-microbe symbiosis may lead to pest-resistant crops

Next   Learn more