An Eidosoma research project

Cells

Modelling how living systems remember.

Cells is our open-source simulator for exploring how molecules assemble, cells communicate, and tissues organise — and how these processes shape memory and inheritance.

Explore on GitHub Read the documentation

Python + JAXCPU + GPUMIT licensed

The research question

What lets a living system remember?

A signal can disappear while its effects persist. We’re building Cells to investigate where that history is held, how it changes future behaviour, and whether it survives when the cells or molecules carrying it are replaced.

The Cells research programme: ask questions from two preprints, model cells and tissues, remove cues and disrupt pathways, then measure memory, inheritance and form. — The aim is to test our ideas in increasingly realistic simulations, including experiments that remove signals, interrupt communication, and challenge cells with different histories.

Plastic Heredity

When does inherited organisation break, recover, or persist? What allows an acquired state to pass to new carriers and be renewed?

Read the introduction

How Patterns Return

How do experience, retained cellular state, and tissue organisation affect which patterns and functions can be expressed, repaired, or rebuilt?

Read the preprint

Cells today

From molecular assembly to communicating cells.

The simulator brings together several model families, with explicit equations, numerical checks, and a record of what each model has been verified to do.

Cells today: bioelectricity, calcium signalling and protein assembly, with a shared queue for independent GPU simulations and saved results and checkpoints. CPU and L4 are verified locally; Jupiter qualification is planned. — The GPU workers shown represent independent simulations. Numerical qualification has used CPUs and one NVIDIA L4; running a single interacting tissue across multiple GPUs requires further development.

Bioelectricity: Ion transport, membrane voltage, pumps, channels, gap junctions, and extracellular fields in selected cell and tissue models.
Cell chemistry and regulation: Calcium and IP3 signalling, internal compartments, and chemical reactions, alongside isolated regulatory models and a bounded two-cell regulatory and ionic model.
Protein assembly: Stochastic cage formation, finite material pools, and breakable bonds. Connecting assembly to cell physiology is an experimental modelling hypothesis.
Repeatable experiments: Save a complete state, resume it, or branch into alternative futures. Run independent simulations with CPU references, local GPU comparisons, and conservation checks.

Follow development and verification

Where we’re going

Toward whole cells, tissues, and organs.

We’re working toward richer cell physiology, broader comparisons with the BETSE and BITSEY reference simulators, and experiments that test how cellular history is written, retained, and transmitted.

Planned development: broader BETSE channel and transport coverage, integrated assembly, energy and mechanics, Jupiter scaling from four to sixteen GPUs, and tissue-to-organ models. — These are planned directions. Energy metabolism, mechanics, growth, turnover, and organ models extend the programme. The organ illustration is schematic; a first organ and species have yet to be selected.

Whole-cell models and interacting organ systems are long-term goals. Calibration against measured biology and qualification on Jupiter’s multi-GPU infrastructure remain ahead. Each comparison is tied to a specific model and set of conditions.

Explore the roadmap

Built in the open

Explore the models.
Follow the evidence.

Cells is written in Python and JAX and released under the MIT License. The repository brings together the code, model guides, verification reports, and scientific background.

Visit the repository Discuss a collaboration