Examples

Below is an overview of the examples, organized by model type or analysis type.

• CRM/

  • examples-sim-CRM.ipynb: Simulate time-course data using the Cross-Feeding Resource Model.

• gLV/

  • examples-bayes-gLV.ipynb: Bayesian inference for Generalized Lotka-Volterra.

  • examples-lasso-gLV.ipynb: Lasso-based inference for gLV.

  • examples-ridge-gLV.ipynb: Ridge-based inference for gLV.

  • examples-Rutter-Dekker.ipynb: Rutter-Dekker example.

  • examples-sim-gLV.ipynb: Simulation of gLV dynamics.

  • examples-Stein.ipynb: Real-life dataset example (Stein et al.) applying gLV methods. (Additional CSV files support these notebooks.)

• gMLV/

  Generalized Metabolic Lotka-Volterra (gMLV) is a variation of gLV that includes metabolite interactions alongside microbial abundances.

  • examples-ridge-lasso-gMLV.ipynb: Ridge/Lasso inference for gMLV.

  • examples-sim-gMLV.ipynb: gMLV simulation examples.

• GP/

  • examples-impute-GP.ipynb: Data imputation with Gaussian Processes.

  • examples-impute-GP_Stein.ipynb: Extended GP-based imputation for Stein dataset. (CSV files for input/output data are included here.)

• MultiModel/Herold/

  These notebooks use real-life data from Herold et al. to demonstrate multi-model workflows.

  • examples-Herold-sVAR.ipynb: sVAR approach on Herold dataset.

  • examples-Herold-VAR.ipynb: VAR approach on Herold dataset.

  • examples_impute_data.ipynb: Data imputation for Herold multi-model workflows. (`Source Data/` folder contains all raw files needed for these notebooks.)

• MVAR/

  • examples-infer-MVAR.ipynb: Inference with the Multivariate Autoregressive model.

  • examples-sim-MVAR.ipynb: Simulation using MVAR. (`parametersS.json` is included for these demos.)

• VAR/

  • examples-bayes-VAR.ipynb: Bayesian inference for Vector Autoregression.

  • examples-sim-VAR.ipynb: Simulation examples for VAR. (JSON files and CSV data support these notebooks.)

• run_gMLV_sims.py: A script to run gMLV simulations from the command line.

Each subfolder includes one or more Jupyter notebooks that guide you step-by-step through setup, simulation, and analysis with the MIMIC package. Feel free to explore each subfolder for model-specific usage instructions, parameter files, and example data.

Jupyter Notebook Examples by Model

CRM

• Simulate some time course data from the CRM
  • Model with five species
  • Simulate single time course

GP

• Imputing Missing Values using Gaussian Process
• Using the Data Imputator on a Real Public Dataset - Stein et al. (2013)

MVAR

• MVAR Simulation for Microbiota and Metabolites
  • MVAR Process
  • Objective
  • Simulation with VARsim.py for sVAR Models
  • Example Usage of VARsim.py for MVAR Simulation
• MVAR Simulation for Microbiota and Metabolites
  • MVAR Process
  • Objective
  • Simulation with VARsim.py for sVAR Models
  • Example Usage of VARsim.py for MVAR Simulation
  • Visualization

MultiModel

• VAR inference on a Real Public Dataset, Herold et al. (2020)
• Sparse VAR on a Real Public Dataset, Herold et al. (2020)
• Data imputation on a Real Public Dataset, Herold et al. (2020)
  • Figure 3 - Community Structure and Function Dynamics:
  • Figure 4 - Metabolite Levels and Environmental Dynamics:

VAR

• Bayesian Inference for Vector AutoRegression (VAR) Models
  • Introduction to Bayesian VAR
  • Notebook Structure
  • Objective
  • Example Usage
• Vector AutoRegression (VAR) Simulation
  • Introduction
  • Notebook Structure
  • Objective

gLV

• Repeat Rutter & Dekker et al 2024 analysis
• Perform Bayesian inference with shrinkage
• Repeat Stein et al. 2013 analysis
• Used Bayesian inference to infer the parameters of a (linearised) gLV model
  • Bayesian inference with no shrinkage
  • Bayesian inference with shrinkage: Horseshoe prior
  • Bayesian inference with shrinkage and a perturbation with unknown interactions
• Example Lasso gLV model
• Simulate some time course data and perform ridge regression as in Stein et al. 2013
  • Single time course
  • Multiple time course
  • Multiple time course including perturbations
• Simulate some time course data from the gLV
  • Model with five species
  • Simulate single time course
  • Simulate single time course with a perturbation
  • Simulate multiple time courses
  • Simulate multiple time courses with a perturbation

gMLV

• Run ridge regression and lasso for the gMLV model
  • Five species, six metabolites, single time course
  • Perform ridge regression to get the abundance model
  • Perform lasso regression to obtain interactions between X and S
• Simulate some time course data from the gMLV
  • Five species, six metabolites, single time course
  • Five species, single time course, with a perturbation