Daphnia-tutorial
PanelPOMP Data Analysis Tutorial — Supplement 2 for Mechanistic Models for Panel Data
Tutorial for PanelPOMP Data Analysis
Supplement 2 for “Mechanistic Models for Panel Data: Analysis of Ecological Experiments with Four Interacting Species”
Bo Yang, Jesse Wheeler, Meghan A. Duffy, Aaron A. King, Edward L. Ionides
| Preprint: arXiv:2506.04508 | Archive:  |
→ Open the Tutorial
Overview
This tutorial provides step-by-step guidance for implementing Panel Iterated Filtering (PIF) methods using the panelPomp R package. It accompanies the main analysis in Mechanistic models for panel data: Analysis of ecological experiments with four interacting species.
The tutorial walks through two complete worked examples — from raw data and model specification through parameter estimation, diagnostics, and profile likelihood confidence intervals — using the Daphnia mesocosm experimental system.
Tutorial Contents
The tutorial (tut.html) covers two models of increasing complexity:
Section 1: SRJF Model (Single Species, No Parasite)
A simpler entry point with 3 latent states (Susceptible, Juvenile, Food):
- Experimental design and data structure — loading and exploring the mesocosm data
- Mechanistic model — stochastic differential equations, biological mechanisms, parameter definitions
- PanelPOMP implementation — process model simulator (
Csnippet), initial states, measurement model - Parameter estimation via PIF — multi-stage optimization with tempering
- Diagnostics:
- Parameter scaling verification
- Evidence for unit-specific parameterization
- MIF search visualization (convergence diagnostics)
- Monte Carlo Adjusted Profile (MCAP) confidence intervals
Section 2: SIRJPF2 Model (Two Species + Parasite)
The flagship model with 8 latent states:
- Experimental design — mixed-species parasitized treatment
- Mechanistic model — full two-species SDE system with infection and parasite dynamics
- Parameter estimation with block sampling — handling high-dimensional parameter spaces
Prerequisites
Before running the tutorial, install the required R packages:
install.packages(c("pomp", "panelPomp", "tidyverse"))
Computational Requirements
The tutorial includes computationally intensive examples controlled by run_level:
| Level | Purpose | Approximate Resources |
|---|---|---|
| 1 | Quick debugging | Minutes, any machine |
| 2 | Local computation | ~1 hour on 36 cores, ~50 GB memory |
| 3 | Full estimation | ~40 hours on 36 cores, ~50 GB memory |
The rendered tutorial (tut.html) includes all pre-computed outputs so you can read through the results without running the code.
Source Files
| File | Description |
|---|---|
tut.qmd |
Quarto source for the tutorial |
tut.html |
Rendered tutorial (pre-computed results) |
data/ |
Data files used in the tutorial |
bib-daphnia.bib |
Bibliography |
Makefile |
Build automation |
Render the tutorial locally:
make
or:
Rscript render_tut.R
Related Resources
| Resource | Link |
|---|---|
| Main analysis repository | github.com/Megumi-ybb/Daphnia-ms |
| Main analysis website | megumi-ybb.github.io/Daphnia-ms |
| panelPomp R package | CRAN |
| pomp R package | kingaa.github.io/pomp |
| Zenodo archive | doi:10.5281/zenodo.15595669 |
Citation
Yang, B., Wheeler, J., Duffy, M. A., King, A. A., & Ionides, E. L. Mechanistic models for panel data: Analysis of ecological experiments with four interacting species. arXiv:2506.04508