Synthetic Biology#
Tutorials for the omicverse.synbio module — a self-contained three-layer
design stack that bridges metabolism, protein / enzyme engineering,
and DNA.
A synthetic-biology project rarely lives in one layer. You reason about a
pathway’s yield (metabolism), the enzymes that carry its flux (protein),
and the genes you will actually build (DNA). ov.synbio puts all three in one
place — and, crucially, wires them together. Its signature is the A↔B hinge:
predict a turnover number from an enzyme’s sequence, push it into a
genome-scale metabolic model as an enzyme-capacity constraint, and re-solve the
attainable yield. Edit the enzyme → the metabolic network re-solves.
Layer |
Functions |
|---|---|
A — metabolic (COBRApy) |
|
B — protein/enzyme (ESM / ProteinMPNN) |
|
C — DNA (DNAchisel / primer3) |
|
The introductory notebook runs all three layers on real data — the
e_coli_core genome-scale model, the 56-residue GB1 domain, and E. coli
phosphofructokinase — and closes with the A↔B coupling.
- Synthetic biology with
ov.synbio— from metabolism to enzyme to DNA - ov.synbio II — circuits, CRISPR, assembly & pathway design
- ov.synbio III — CRISPR editing & directed-evolution libraries
- ov.synbio IV — advanced SOTA: mRNA design, de-novo binders, prime editing
- Case study I — an E. coli succinate cell factory
- Case study II — engineering a more thermostable DHFR
- Case study III — evaluating a protein design (does it get better?)
Installation#
pip install 'omicverse[synbio]'
The metabolic and DNA layers are CPU-only (COBRApy with the free GLPK solver,
DNAchisel, primer3). The protein layer uses a GPU automatically when available
and reuses omicverse’s existing PyTorch dependency; ESM / ESMFold / ProteinMPNN
weights download on first use to ~/.omicverse/synbio_weights (override with
OMICOS_SYNBIO_WEIGHTS). Every backend is optional and gated behind an
actionable error — import omicverse never requires any of them.