Stock cross-check
Compares the plan against your own lab's inventory — what's already on the shelf, what needs ordering, and from where.
For resource-constrained fly labs
Design a fly cross in an afternoon, not a month.
Synapsis turns a research goal into a validated, generation-by-generation plan — balancers tracked, markers chosen, lethal combinations caught before they cost eight weeks on the bench.
~3 weeks
spent planning the cross
1 afternoon
same scheme, fully validated
Bench time for the worked example below is about three weeks either way — that's fly biology, not planning. What the copilot compresses is everything that happens before the first cross is even set up.
How it works
The model explains the scheme. It doesn’t invent it.
A wrong cross is invisible until eight weeks in — this is the one place on the page worth being explicit about the architecture.
- 01
You describe the goal
Plain language, the way you'd explain it to a labmate: “knock down parkin in dopaminergic neurons and image mitochondrial morphology.”
- 02
A solver builds the scheme
A deterministic constraint solver, not a language model, reasons over chromosomes, balancers, and markers to generate the cross tree. Genetics is math here, not a guess.
- 03
The plan is explained back to you
The model's only job is translation: turning the solver's output into an annotated, plain-language pedigree you can hand to a bench scientist or a reviewing PI.
Worked example
One real goal, planned start to finish.
This example is real DRTC inventory data, run through an illustrative gene target. Join the waitlist and the pilot runs this on your goal, against your own stock list.
Research goal
> knock down parkin in dopaminergic neurons and image mitochondrial morphology
Available in DRTC inventory
- ✓ w[1118]; TH-GAL4/TM6B
- ✓ w[1118]; +/CyO
- ✕ UAS-Parkin-RNAi — order from Bloomington
Generated scheme
- P0+2d
♀ w[1118]; UAS-Parkin-RNAi/CyO × ♂ w[1118]; TH-GAL4/TM6B
Set the cross — no selection yet
- F1+10d
F1 progeny of P0
Select non-Curly, non-Tubby: UAS-Parkin-RNAi/TH-GAL4
- F2+10d
F1 siblings crossed, balanced over CyO
Select Curly-winged: UAS-Parkin-RNAi/CyO ; TH-GAL4
- Result · ~22 days
w[1118]; UAS-Parkin-RNAi/CyO ; TH-GAL4 — stable stock
Copy this scheme as plain text
- P0 (+2 days): cross ♀ w[1118]; UAS-Parkin-RNAi/CyO × ♂ w[1118]; TH-GAL4/TM6B. Set the cross — no selection yet.
- F1 (+10 days): cross F1 progeny of P0. Select non-Curly, non-Tubby: UAS-Parkin-RNAi/TH-GAL4.
- F2 (+10 days): cross F1 siblings crossed, balanced over CyO. Select Curly-winged: UAS-Parkin-RNAi/CyO ; TH-GAL4.
- Result (~22 days total): w[1118]; UAS-Parkin-RNAi/CyO ; TH-GAL4 — stable stock.
What it handles
Balancer & marker reasoning
Works out which visible marker to select on at each generation, and flags lethal combinations before you set a cross.
Timeline estimation
Generation-by-generation dates based on your lab's own rearing conditions, not a generic textbook average.
Exportable bench sheet
A printable plan for the bench — not a chat transcript you have to re-read and transcribe.
Built for two readers
A plan the postgrad can run and the PI can trust.
For the researcher
A working, step-by-step plan you can start executing today — no more sketching pedigrees on paper or re-deriving marker logic from a textbook you half-remember.
For the PI
An auditable plan — every selection step comes from the same deterministic solver every time, with its reasoning shown, not a model improvising and hoping the genetics holds up.
Where this actually stands
This is a work in progress, not a finished product. Everything on this page describes what Synapsis would do — the concept, the solver-plus-explainer architecture, and one real worked example. Joining the waitlist gets you early access when the pilot opens, built first around a real cross scheme, not a demo.
Be first in line when the pilot opens.
No spam, no drip campaign — one email, when there's a working pilot to try, built around a real cross scheme.
Prefer email? Reach out directly at hello@synapsis.org.uk