Product Overview

Biosimulant is building a product for discovering, running, and composing biological simulations with a modern UX.

This documentation is public-facing. It intentionally avoids internal implementation details.

Vision

No one currently owns “run any biology simulation with a unified UX.” Biosimulant fills this gap by:

• Aggregating existing tools and databases (not competing with them)
• Composition for multi-model, multi-scale simulations
• Modern browser-based UX vs. desktop/dated alternatives
• Local-first + cloud-ready workflow

What is bsim?

The product has two layers:

B-Simulant Core Library

The public Python library (pip install bsim) providing:

• Composable Runtime - Event-driven simulation orchestration
• BioModule System - Pluggable simulation components with ports and signals
• Adapters - Wrap external simulators/standards as modules
• Domain Packs - Pre-built modules for specific domains (e.g. neuroscience, ecology)
• SimUI - Python-declared web UI for local runs

bsim Platform

A product layer that aims to provide:

• Model Repository - Aggregated models from BioModels, ModelDB, NeuroML-DB, CellML
• Execution - Run simulations with a consistent run/result experience
• Composition Editor - Visual drag-and-drop model composition
• Project Management - Organize runs and share with collaborators

Supported Standards

Model Sources

Key Features

Model Discovery

• Search and filter across models and metadata
• Standards-aware views (e.g. SBML vs NeuroML vs ML)
• Reproducible runs (config + parameters + outputs)

Simulation Execution

• Configurable parameters (duration, dt, initial conditions)
• Multiple backends via adapters (SBML, NeuroML, ML, etc.)
• Consistent results format for analysis and visualization

Composable Simulations

Build complex multi-scale models with the B-Simulant runtime:

from bsim.adapters import TelluriumAdapter, MLAdapter, TimeBroker
 
sbml = TelluriumAdapter(model_path="glycolysis.xml", expose=["ATP"])
ml = MLAdapter(model_path="drug_response.onnx", inputs={"ATP": "x1"}, outputs={"y": "efficacy"})
 
broker = TimeBroker()
broker.register("metabolism", sbml, time_scale="seconds")
broker.register("predictor", ml, time_scale="seconds")
broker.connect("metabolism.ATP", "predictor.ATP")
broker.setup()
 
for _t in broker.run(duration=100.0, dt=0.1):
    pass

Domain Packs

Pre-built modules for specific domains:

Neuroscience Pack (bsim.packs.neuro):

• IzhikevichPopulation - Spiking neurons with presets (RS, FS, Bursting)
• PoissonInput - Spike train generation
• ExpSynapseCurrent - Synaptic integration
• SpikeMonitor, RateMonitor, StateMonitor - Visualization

Ecology Pack (bsim.packs.ecology):

• OrganismPopulation - Population dynamics
• PredatorPreyInteraction - Lotka-Volterra
• CompetitionInteraction, MutualismInteraction
• PopulationMonitor, EcologyMetrics

Visualization

• Time series plots for species concentrations
• Raster plots for spike trains
• Phase space for ecological dynamics
• JSON export with full metadata

Who is bsim For?

Computational Biologists

• Run SBML models without installing tellurium locally
• Access BioModels database through unified interface
• Compose multi-model simulations

Neuroscientists

• Simulate spiking neural networks
• Use NeuroML models from NeuroML-DB
• Visualize spike rasters and firing rates

Systems Biologists

• Explore metabolic pathways
• Parameter sensitivity analysis
• Hybrid mechanistic + ML models

Educators & Students

• Interactive Lotka-Volterra predator-prey
• Hodgkin-Huxley action potentials
• No installation required

Getting Started