Skip to content

Latest commit

 

History

History
368 lines (291 loc) · 13.5 KB

README.md

File metadata and controls

368 lines (291 loc) · 13.5 KB

Analysis Code for Whole-Cell E. coli Colony Simulations

This repository contains code to accompany a paper "Whole-Colony Modeling of Escherichia coli" that is currently in preparation. The associated data are available under the DOI 10.5281/zenodo.4697519 and can be re-generated using the model simulation code with DOI 10.5281/zenodo.4695018. This project is hosted on GitHub at https://github.com/CovertLab/wcecoli-colony-analysis and licensed under the terms in LICENSE.txt.

We welcome comments, which you can leave by filing an issue on the GitHub repository.

Requirements

Hardware Requirements

  • 5 GB of free space because the un-compressed simulation data is about 2.4 GB.
  • At least 4 GB RAM because you will need to load at least one experiment's data into memory to generate the figures. This data can be over 1 GB. If you generate all the figures at once, the analysis code will need to load all 2.4 GB of simulation data into memory. Depending on what other processes consume RAM on your system, you may need more memory and/or swap space.

Software Requirements

  • If you are running on a headless system, you will need to install the X virtual framebuffer (XVFB) and prepend the figure generation commands with xvfb-run -a.
  • To analyze phylogenetic data, you will need to have R (version 4.0.4) with the argparse and phytools packages. Developers will also need the lintr package for linting.
  • Python 3.8.3
  • If you clone from GitHub, you'll also need these tools:
  • GPG to verify code integrity if you don't want to rely on GitHub's security checking.

Setup

  1. Clone the repository.

    $ git clone https://github.com/CovertLab/wcecoli-colony-analysis.git

    Alternatively, you can simply extract an archive of the source code if you got the code in that format.

    You should always validate the code's integrity and make sure it comes from a trusted source. See the Security section below for instructions.

  2. (recommended but optional) Setup a Python virtual environment.

  3. Install Python dependencies

    $ pip install numpy
    $ pip install -r requirements.txt

    Note that you have to install numpy first because the setup.py script of one of our dependencies requires it.

  4. Install R dependencies. In an R shell, run:

    install.packages('argparse')
    install.packages('phytools')

    If you want to run the lint checks (only developers need to do this), also run:

    install.packages('lintr')

Reproduce Analyses

There are two ways to make the simulation data accessible to the analysis code: reading from files or reading from a MongoDB database. Here, we will only describe how to read from files since MongoDB is over-kill for just reproducing our analyses.

  1. The raw simulation data is not included in the repository due to its size. Instead, it is available under the DOI 10.5281/zenodo.4697519. Follow the DOI link and download all the files that it contains to data/.

    You should verify the integrity of the downloaded files. You can do so like this:

    $ cd data
    $ gpg --verify SHA512SUMS.txt.asc
    gpg: assuming signed data in 'SHA512SUMS.txt'
    gpg: Signature made Fri Apr 16 14:26:12 2021 EDT
    gpg:                using RSA key F76925D5D12B91104587678FC98CBB9C501917E0
    gpg: Good signature from "anonymous <cs.temporary@icloud.com>" [full]
    $ shasum -c SHA512SUMS.txt
    LICENSE.txt: OK
    README.md: OK
    archived_simulations.tar.gz: OK
    search.json: OK
  2. The raw simulation data is now an archive at data/archived_simulations.tar.gz. You can extract it like this:

    $ cd data
    $ tar -xf archived_simulations.tar.gz

    Now you should see the simulation data as a series of JSON files under data/archived_simulations.

    We also provide the data/search.json file, which specifies the theoretical boundary that appears in magenta in Figures 5G 5H, and 5I. This is already un-compressed, so you don't need to do anything more with it except pass it to scripts as specified below.

  3. Now you can generate figures using the src/make_figures.py script. Run python -m src.make_figures -h to see the available options:

    $ python -m src.make_figures -h
    usage: make_figures.py [-h] [--atlas] [--port PORT]
                           [--host HOST]
                           [--database_name DATABASE_NAME]
                           [--data_path DATA_PATH] [--3A] [--3B]
                           [--3C] [--3D] [--3E] [--3F] [--3G]
                           [--5A] [--5B] [--5C] [--5D] [--5E]
                           [--5F] [--5G] [--5H] [--5I] [--X1]
                           [--all]
                           search_data
    
    Generate selected figures and associated stats from
    simulation data.
    
    positional arguments:
      search_data           Path to boundary search data.
    
    optional arguments:
      -h, --help            show this help message and exit
      --atlas, -a           Read data from an mongoDB Atlas
                            instead of a local mongoDB.
                            Credentials, cluster subdomain, and
                            database name should be specified in
                            secrets.json.
      --port PORT, -p PORT  Port at which to access local mongoDB
                            instance. Defaults to "27017".
      --host HOST, -o HOST  Host at which to access local mongoDB
                            instance. Defaults to "localhost".
      --database_name DATABASE_NAME, -d DATABASE_NAME
                            Name of database on local mongoDB
                            instance to read from. Defaults to
                            "simulations".
      --data_path DATA_PATH
                            Folder of JSON files to read data
                            from instead of Mongo
      --3A                  Generate figure & stats for fig 3A:
                            snapshots of growing colony consuming
                            glucose
      --3B                  Generate figure & stats for fig 3B:
                            environment cross-sections showing
                            glucose depletion
      --3C                  Generate figure & stats for fig 3C:
                            snapshots in the basal condition
      --3D                  Generate figure & stats for fig 3D:
                            snapshots in the anaerobic condition
      --3E                  Generate figure & stats for fig 3E:
                            colony mass on basal and anaerobic
                            media
      --3F                  Generate figure & stats for fig 3F:
                            snapshots showing expression
                            heterogeneity
      --3G                  Generate figure & stats for fig 3G:
                            distributions of protein
                            concentrations
      --5A                  Generate figure & stats for fig 5A:
                            parameter scan for tolerance
                            threshold
      --5B                  Generate figure & stats for fig 5B:
                            snapshot of final colony under
                            nitrocefin
      --5C                  Generate figure & stats for fig 5C:
                            box plot showing distances from
                            center
      --5D                  Generate figure & stats for fig 5D:
                            phylogenetic tree
      --5E                  Generate figure & stats for fig 5E:
                            dotplot of final [AmpC] colored by
                            survival
      --5F                  Generate figure & stats for fig 5F:
                            dotplot of final [AcrAB-TolC] colored
                            by survival
      --5G                  Generate figure & stats for fig 5G:
                            final [AmpC] and plotted against
                            [AcrAB-TolC]
      --5H                  Generate figure & stats for fig 5H:
                            paths of dead agents through
                            concentration space
      --5I                  Generate figure & stats for fig 5I:
                            paths of a lineage of agents through
                            concentration space
      --X1                  Generate figure & stats for fig X1:
                            death_snapshots_antibiotic
      --all                 Generate all figures and stats.

    For example, to generate Figure 3A from the paper:

    $ python -m src.make_figures data/search.json --data_path data/archived_simulations --3A

    This will save the plots used in Figure 3A to out/figs/. Note that in this case there will be multiple plots generated, one for each simulation, even though only one of these plots was actually used for Figure 3A in the paper. Along with the figures, a stats.json file will also be created.

    You can also generate figures and associated statistics for all the figures, even those that don't appear in the paper (e.g. X1). You can do this like so:

    $ python -m src.make_figures data/search.json --data_path data/archived_simulations --all
  4. Calculate summary statistics using src/analyze_stats.py. You can view its help text by running:

    $ python -m src.analyze_stats -h
    usage: analyze_stats.py [-h] [-o OUT] stats_json
    
    positional arguments:
      stats_json         Path to stats JSON file
    
    optional arguments:
      -h, --help         show this help message and exit
      -o OUT, --out OUT  Path to write summary stats to

    For example, you could generate summary statistics from the figure 3A statistics by running:

    $ python -m src.analyze_stats out/figs/stats.json -o out/figs/summary_stats.json

    The out/figs/summary_stats.json file stores the summary statistics in a human-readable format.

  5. Analyze phylogeny data using src/analyze_phylogeny.r like this:

    $ Rscript src/analyze_phylogeny.r out/figs/phylogeny.nw out/figs/agent_survival.csv

    The analysis will be printed to the console.

For Developers

Create Simulation Data Archives

You can use the src/archive_experiments.py script to create an archive of all the simulation data used by src/make_figures.py:

$ python -m src.archive_experiments <connection args>

where <connection args> can include -o IP for MongoDB IP address IP and -p PORT for MongoDB port PORT.

Run Tests

We use mypy for type checking, pylint for linting, and pytest for unit tests. You can run all these by executing the test.sh script. There should be no errors.

Security

This code is not hardened against malicious inputs. Therefore, you should only run it on data you trust is not malformed. This trusted data might include the archived simulation data we provide or the outputs of simulations you ran.

All releases and commits are signed with the OpenPGP key 0x12C1B5FF558317E5, which has the fingerprint:

1642C5C9C092F5AC1FE9222012C1B5FF558317E5

Verifying Release Signatures

You can validate releases by verifying tag signatures:

$ git verify-tag <tag name>

If you retrieved an archive of the code with an accompanying signature, you can also verify that signature:

$ gpg --verify <signature file>

Verifying Commit Signatures

You can use git log --show-signature to verify these signatures, though note that the signatures will show a fingerprint of the signing sub-key:

F76925D5D12B91104587678FC98CBB9C501917E0

We have provided a script to automatically do this check in check_signatures.py, though of course you should make sure you understand the script before you trust it.

Verifying Using GitHub

If you trust GitHub, you may also rely on the Verified labels GitHub adds to signed commits and releases. Click on the label to check which key was used.

Acknowledgments

This work was supported by the Paul G. Allen Frontiers Group via the Allen Discovery Center at Stanford and NIGMS of the National Institutes of Health under award number F32GM137464. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

This program's development was supported by Eran Agmon and Ryan Spangler, and it was overseen by Markus Covert at Stanford University.

License

This project is licensed under the GNU General Public License (GPL) because it has GPL-licensed dependencies.

Copyright (C) 2021 Christopher Skalnik

This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program. If not, see https://www.gnu.org/licenses/.