Genetic Circuit
Design a construct that computes: give it a logic function and it picks characterized repressor gates, predicts the dynamic range between ON and OFF, and emits the DNA.
| Reached by | Tools → Genetic Circuit |
| Needs | Nothing open — a circuit is designed from a logic function, not from a sequence |
| Formats | FASTA out (one record per device of the construct) |
What it does
Most design work tunes how strongly a gene is expressed. A circuit is a different question: a construct that senses something, applies a logic function to what it sees, and drives a reporter. "Fluoresce when arabinose is present but IPTG is not" is a circuit, and it is built out of repressors wired so that one gate's output promoter is the next gate's input.
The logic comes first. You give it the function — a named gate, or a truth table written as one
bit per input combination with the first input most significant, so AND is 0001. It then builds
the smallest NOR/NOT network that computes it. The gate set is NOR and NOT because that is what
repression naturally gives you: a repressor turns its promoter OFF when it is present, which is an
inverter. An AND therefore costs three gates and an exclusive-OR costs six, and that cost is the
real constraint — every gate in the network has to be a different repressor, or two gates in the
same circuit would silence each other.
Then the parts. A characterized gate library gives each repressor a measured response: how much output it makes when fully off, how much when fully on, the input level at which it switches, and how sharply. Those four numbers are measurements from published work, not estimates made here. The tool searches every way of assigning the library's sensors, repressors and reporters onto your network, scores each one, and keeps the best.
That search is the part that matters most, and it is easy to underestimate. On one library's AND circuit, the worst assignment scores 1.0 — the output is the same whether the circuit is ON or OFF, so it is not a switch at all — while the best scores 19.6. Same logic, same library, same parts list: only the pairing differs. The search is exhaustive rather than sampled, so when it reports a best it is the best, and when the space is too large to enumerate it says so and asks you to narrow it rather than quietly checking a fraction.
Finally the DNA. Each device is assembled from the library's own part sequences in the order its structure declares: a self-cleaving ribozyme, a ribosome binding site, the repressor's coding sequence, and a terminator. The ribozyme is what makes the prediction meaningful — it cuts the transcript at a fixed point so the gate's behaviour does not depend on which promoter is feeding it, which is the property that lets gates be characterized once and then composed.
What this computes
Per circuit
| Readout | What it is |
|---|---|
| Dynamic range | The dimmest ON output divided by the brightest OFF output, across every input combination. The number a decision to build or not build turns on. Reported per output device, and the circuit's own figure is the worst of them — a circuit is only as good as its weakest readout. |
| Truth table | Every input combination, with both the logic value the circuit intends and the level each node actually reaches. The two can disagree, and where they do is where the design fails. |
| Gate assignment | Which sensor, repressor and reporter each node was given, how many assignments were searched, and how many tied for best. A tie means the choice was not forced, which is worth knowing before ordering DNA. |
Per node, per input combination
| Readout | What it is |
|---|---|
| Output level in RPU | Relative Promoter Units — expression measured against the library's own reference standard, which is what makes numbers from different labs comparable at all. |
| Which physical gate won | A repressor family can hold several characterized variants. The one that scored best is named, per row, so a family whose best member changes between rows is visible rather than hidden — that combination is not achievable by any single construct. |
Per construct
| Readout | What it is |
|---|---|
| Length and device list | Each sensor, gate and reporter with its own sequence and its parts named by type. |
| Part inventory | The distinct parts used, so the construct can be read as a bill of materials rather than a wall of bases. |
| Device order along the construct | The arrangement the library's own placement rules put the devices in: which comes first, which pairs must be adjacent, which must precede which, and where each sits relative to the landing pads the library declares. |
| Constraints applied | How many of the library's placement rules bound this construct, out of every rule it declares. Most rules in a library are about devices a given design does not use; the count says which ones actually decided this arrangement. |
The estimate's own conditions
| Readout | What it is |
|---|---|
| Host, medium, temperature and protocol | The conditions every parameter in the library was measured under, named beside the prediction. A predicted dynamic range inherits the conditions of the measurements behind it, and a number shown without them invites an experiment that was never going to match. |
| What the prediction assumes | Stated with every result: insulated parts, no metabolic burden or competition for host resources, steady state, and that growth condition. |
Controls
| Control | Default | What it sets |
|---|---|---|
| Library | Eco1C1G1T1 | The characterized gate library the search draws from |
| Sensors · Gate groups · Reporters | the library's own | Which parts the search may use. The available names are listed under each box |
| Assignment budget | 200,000 | The search is exhaustive, so it is never sampled: past this many assignments it refuses and says what to narrow |
What it does not do
It does not model burden. Gate behaviour measured in isolation is not gate behaviour in a circuit competing for the same ribosomes and polymerases. The prediction assumes no load, which is optimistic in the direction that costs an experiment, and it says so.
It does not invent a part. A repressor with no measured response is not in the library and cannot be assigned. A short library with real provenance is more useful than a padded one.
Provenance of correctness
The gate parameters, part sequences and placement rules are the published Cello libraries, used under Apache 2.0. The device order is solved against those rules and checked against miniEugene, the constraint solver Cello itself uses, over every shipped library's real rule set and over arrangements it labelled valid and invalid one by one. Every stage is checked against Cello itself: the library reader against its own parser over every record of all six libraries, the gate response against its evaluator over 600 swept points, the circuit score against its scorer over 24 circuits, and the assignment search against its own exhaustive search over 2,160 assignments.
CRISPR Guide Design
Find the guides in a target sequence and check where else they could cut.
In-Situ Probes
Design a hybridization probe against a transcript or a locus, padlock probes for in-situ sequencing, or HCR v3.0 split-initiator probe pairs: tile, screen for off-target binding, and pick what images.