Embedded Files
Guiding metabolic engineering via multiomics data and machine learning
Guiding metabolic engineering via multiomics data and machine learning
We will show how the combination of tools presented here (ICE + EDD + ART + Jupyter notebooks) provide a standardized manner to store data so it can be leveraged to produce actionable recommendations and guide synthetic biology.
ICE is an open source repository platform for managing information about DNA parts and plasmids, proteins, microbial host strains, and plant seeds. EDD is an open source online repository of experimental data and metadata. ART is a library that leverages machine learning for synthetic biology purposes, providing predictive models and recommendations for the next set of experiments.
When combined, this set of tools can effectively store, visualize, and leverage synthetic biology data to enable predictive bioengineering and effective actionable items for the next DBTL cycle.
You can find more information in this paper.
We start with a base strain (wild type, or WT) that is bioengineered to produce isoprenol (3-methyl-3-buten-1-ol ). This strain is bioengineered according to several designs (i.e.: knockout malate dehydrogenase, overexpress citrate synthase) suggested by ART. The results are 95 bioengineered strains (BE1,BE2… etc) for which experimental data (isoprenol production levels) are simulated through the Omics Mock Generator library (OMG) and stored in EDD and ICE. These data are then leveraged by ART to recommend, using machine learning, new designs that are expected to improve isoprenol production (REC1, REC2, …). These recommendations and production predictions are compared with the ground truth provided by OMG.
Each of these steps (in orange) is demonstrated through screencasts and Jupyter notebooks, as shown in the table below.
Step by step tutorial
Step by step tutorial
This table provides a guide for all steps in the simulated engineering process described in the figure above.
The process is fully showcased through a combination of Jupyter notebooks and screencasts which can be found below under each step number
Running the full tutorial requires you to download the following libraries from github:
You can download these libraries in a JupyterHub server provided by JBEI, ABF or your own computer.
Step 1
Step 1
Importing the WT strain information into ICE (the public ABF repository).
No files needed for this tutorial.
Screencast1_ICEsingleImport.mp4Step 2
Step 2
Creating the omics data for the WT strain by leveraring the OMG library.
This tutorial requires:
Installing the OMG library from github.
Downloading the multiomics files from github.
Step 3
Step 3
Importing the WT data into EDD.
Required files:
Screencast2_EDDimport.mp4Step 4
Step 4
Visualizing data through EDD
No files needed for this tutorial.
Screencast3_EDDvisualization.mp4Step 5
Step 5
Generating initial design suggestions through ART. These are the modifications of the original strain from step 1.
This tutorial requires:
Installing the ART library from github (academic and commercial licenses are available upon request).
Screencast4 _ICEmultipleImport.mp4Step 7
Step 7
ICE export of bioengineered strains.
No files needed for this tutorial.
Screencast5_ICEexport.mp4Step 8
Step 8
Generating isoprenol production data for the bioengineered strains.
This tutorial requires:
Installing the OMG library from github.
Downloading the multiomics files from github.
Step 9
Step 9
Importing into EDD of bioengineered strains production data (including quick visualization).
You will need these files:
Screencast6_BEupload.mp4Step 10
Step 10
Export of isoprenol production data for bioengineered strains from EDD.
This tutorial requires:
Screencast7_EDDexport.mp4Step 11
Step 11
Training ART and producing recommendations.
This tutorial requires:
Installing the ART library from github (academic and commercial licenses are available upon request).
Screencast8_ARTrecommendations.mp4Step 12
Step 12
Using the ART frontend.
This step requires:
If you don't have an @lbl.gov email, sign up and create an account at art.agilebiofoundry.org
Screencast9_ARTfrontend.mp4Step 13
Step 13
Comparing ART predictions with ground truth.
This tutorial requires:
Downloading the multiomics files from github.
For more information, please email ese-robotics@lbl.gov
Page updated
Report abuse