Passion projects (extra)

Personal passion areas

Morpheus: Bio-inspired voxel cell morpohology simulator https://github.com/liamzebedee/morpheus

Rewriting OpenTron (biolab automation) protocols to be well-engineered, composable typed Python. https://github.com/liamzebedee/Slowpoke-v2

Paper: Bringing the genetically minimal cell to life on a computer in 4D - JVCI-syn3A simulation

A Rust implementation of Radhika Nagpal’s MIT thesis, “Programmable Self-Assembly: Constructing Global Shape using Biologically-inspired Local Interactions and Origami Mathematics” https://github.com/liamzebedee/biogami

Morphology - final presentation

Cigars and tomatoes are both natural products, so in principle you should be able to grow a cigar and pick it off a tree — we just don’t know how to program a cell to grow into that shape. To study this I built an in silico 3D cell simulation that runs a programming language mapping onto real cell mechanics: hormones for signaling, internal state analogous to internal chemical concentrations, logic as found in GRN circuits, neighbour sensing, and the ability to divide. Using an LLM to search this language spec, I discovered a program that grows a single cell into a cylinder. The core idea is how global form arises from purely local interactions — every cell runs the same program with only a local view, no eyes — and it turns out it can be remarkably simple: grow out and grow up. A center cell emits a hormone whose concentration only reaches a certain radius, and cells are programmed to divide when the concentration exceeds a threshold (growing out). A second process grows up: cells inherit a biased chemical state carrying a signal that decays as the structure grows upward, encoding a preset height. There is still far more to discover — we still cannot take a plant, load it into a simulator, alter its developmental program, and reliably grow a new morphology in silico. Understanding how organisms decode and generate form from DNA is likely to be as foundational this century as molecules and atoms were in the early 20th, and now is the perfect time to figure it out.

Readings

The Genomic Code: The genome instantiates a generative model of the organism https://arxiv.org/pdf/2407.15908

Physics of Computation and Behavior in Plants https://arxiv.org/pdf/2604.21763v1

Computational modeling of plant root development: the art and the science

Multi-level gene regulatory network models to understand complex mechanisms underlying plant development https://pubmed.ncbi.nlm.nih.gov/33171396/

cells are like programmable agents that serve as units for a rule-based behavior for the self-assembly of larger multicellular entities: tissues, organs, and organisms. In fact, a portion of the genes expressed in the context of a stable state S are cell-cell communication proteins, such as cytokines and extracellular matrix components, which mediate the interaction between cells to form a higher-level cell-cell interaction network. This cellular network in turn has a state space in which attractor states would represent tissue states, such as inflammatory states, regenerative states, and “tumor states.”

Multistability and Multicellularity: Cell Fates as High-Dimensional Attractors of Gene Regulatory Networks https://www.sciencedirect.com/science/chapter/edited-volume/abs/pii/B9780120887866500332