HRA Hackathon

Simulation of Human Reference Atlas-relevant biological models using COPASI via process-bigraph composites. Models are fetched from BioModels, simulated as time-course experiments, and rendered into an interactive HTML report with species name resolution from UniProt and KEGG.

Results

View all simulation results — interactive Plotly plots with model metadata, hover tooltips, and toggle-able species traces. Clone the repo and open the HTML file in your browser, or click the link above then use GitHub's download button (⬇).

Model ID	Description
BIOMD0000000356	Nyman2011 — Hierarchical insulin-glucose dynamics
BIOMD0000000620	Palmer2014 — IL-1β-blocking therapies in T2DM (disease)
BIOMD0000000621	Palmer2014 — IL-1β-blocking therapies in T2DM (healthy)
BIOMD0000000833	DiCamillo2016 — Insulin signalling pathway
BIOMD0000000854	Gray2016 — The Akt switch model
MODEL1209260000	Koenig2012 — Hepatic glucose metabolism
MODEL1912090001	Huang2014 — Insulin signaling via IRS1 and IRS2
MODEL2401110001	Pancreas glucose model

Setup

Requires Python 3.11–3.12 and uv.

uv sync

Usage

Run the default set of HRA models:

uv run python -m experiments.run_hra_models

Run specific models by passing their BioModels IDs:

uv run python -m experiments.run_hra_models BIOMD0000000356 BIOMD0000000833

Or call programmatically:

from process_bigraph import allocate_core
from experiments.run_hra_models import run_hra_models

core = allocate_core()
run_hra_models(core, model_ids=["BIOMD0000000356"])

Results are saved to out_hra_models/results.html.

Project structure

experiments/
  run_hra_models.py    # Main orchestration — builds documents, runs simulations
  model_ids.py         # Default list of BioModels IDs
  biomodels_fetch.py   # Fetching SBML/SED-ML from BioModels, parsing SED-ML
  species_names.py     # Species name resolution via UniProt & KEGG REST APIs
  sbml_utils.py        # SBML metadata extraction (time units, compartments, etc.)
  plotting.py          # Plotly figure generation and HTML report assembly
processes/
  copasi_process.py    # CopasiUTCStep — process-bigraph Step wrapping COPASI time-course
models/                # Cached SBML files (populated on first run)
out_hra_models/        # Simulation output — JSON state files and results.html

How it works

1. For each model ID, the pipeline fetches metadata and files from BioModels.
2. SED-ML is parsed to extract simulation duration and number of time points.
3. A process-bigraph composite document is built with a CopasiUTCStep and a RAM emitter.
4. COPASI runs the time-course simulation (with automatic retry using stiffer solvers if needed).
5. Species names are resolved from SBML annotations via UniProt and KEGG REST APIs.
6. All results are combined into a single interactive HTML report.