ADToolbox¶

From raw sequencing reads to a calibrated anaerobic digestion model. ADToolbox connects metagenomics evidence, curated reaction and feed databases, and dynamic ADM1 / e-ADM simulations into one reproducible Python and command-line workflow.

Get started Install View on GitHub

What ADToolbox does¶

Anaerobic digestion models such as ADM1 lump the microbial community into a handful of guilds whose initial biomass you are expected to guess. ADToolbox replaces that guess with measurement. It profiles a community two ways — 16S amplicon reads mapped through GTDB and a curated enzyme-to-reaction database, or shotgun reads searched directly against that database for functional evidence — and turns either into the microbial COD allocation that an extended ADM (e-ADM) model actually needs. From there you can simulate, visualize, and fit the model against experimental data.

ADToolbox architecture: microbiome sequence processing and the curated protein,
reaction, feed, and experiment databases feed the ADM model attributes; an optimization
layer fits parameters, and the toolbox is exposed through a command line interface, a web
application, and a Python API.

Explore the docs¶

Quickstart

Install the package, download the reference databases, and run your first ADM simulation in a few minutes.

Quickstart
Metagenomics pipeline

Turn a table of SRA accessions or FASTQ files into model-ready microbial COD allocations, locally or on Slurm.

Pipeline guide
ADM models

The full input contract, stoichiometry, rate laws, and inhibition terms for ADM1 and e-ADM.

Model reference
Parameter tuning

Fit kinetic parameters to experimental data with SciPy, OpenBox, genetic, or neural-surrogate optimizers.

Tuning guide
Command line interface

Every adtoolbox command, its options, and the files it reads and writes.

CLI reference
Python API

Auto-generated reference for core, adm, configs, optimize, utils, and stats.

API reference

Install¶

pipWith optimizersFrom sourceDocker

pip install adtoolbox

pip install "adtoolbox[optimize]"

git clone https://github.com/chan-csu/ADToolbox.git
cd ADToolbox
pip install -e .

docker run --rm parsaghadermazi/adtoolbox:latest adtoolbox --help

ADToolbox requires Python 3.11 or newer. See Installation for container images, HPC notes, and the external tools each pipeline step needs.

At a glance¶

Module	What it gives you
`core`	Databases, feeds, experiments, and the metagenomics pipeline.
`adm`	ADM1 and e-ADM model construction, ODE integration, and plotting.
`configs`	Path and threshold configuration for every other module.
`optimize`	Parameter estimation and model validation against experiments.
`utils`	FASTA helpers, MMseqs2 wrappers, and Slurm job generation.
`stats`	Distance matrices and scaling for feature tables.

Try it without installing¶

The example notebooks run on Binder, though Escher map visualization is not available there.

Credits¶

ADToolbox is developed in the Chan Lab at Colorado State University.

Contributor	Contact
Parsa Ghadermazi	parsa.ghadermazi@colostate.edu
Ethan Rimelman	rimelman@colostate.edu
Siu Hung Joshua Chan (PI)	joshua.chan@colostate.edu