Solutions
For ag-biotech teams
Trait development runs on season-length feedback loops. Moving the first several iterations into software is the only real way to shorten them — and pathway transfer is where that pays best.
65
host organisms
1,000
reference constructs
14.1 Mbp
designed DNA
Free
feasibility modelling
01
The problem
A trait program’s iteration cost is measured in seasons. Every design that reaches the greenhouse and fails on a reason that was knowable in advance — the host could not fund the pathway, the promoter was wrong for the tissue, the guide had a crowded off-target neighbourhood — consumes a slot that a better design could have used.
Cross-species work makes this sharper. Transferring a pathway into a host that never evolved it means almost every assumption carried over from the donor species needs re-examining, and discovering which ones broke by growing the plant is the most expensive possible way to find out.
02
What the platform does about it
Feasibility before commitment
COBRApy flux analysis against a host's own genome-scale model answers whether the carbon and nitrogen budget supports a target. It runs on CPU, costs nothing, and routinely eliminates designs in an afternoon.
Whole-panel sweeps
Batch mode runs the same analysis across dozens of hosts or candidates with parameters pinned once, so a host-selection decision rests on a comparable table rather than accumulated impressions.
Guides with evidence
CHOPCHOP- and CRISPOR-style scoring proposes sites; Cas-OFFinder sweeps the host genome so off-target context is part of the design record rather than a later surprise.
Tissue-aware choices
The expression atlas puts transcriptomic context behind promoter selection. Accumulation is tissue-specific, and a promoter that performs in the wrong tissue produces nothing anyone can measure.
Packages a pipeline can take
GenBank, FASTA, SBOL, ST.26 listings, IDT and Twist order sheets, Addgene references, verification primers, a validation report and a checksummed manifest — in one versioned archive.
Workspace rules
Teams define required annotations, permitted hosts, forbidden parts and review requirements once; validation enforces them identically across every project in the workspace.
03
A representative project
- Program
- Cross-species pathway transfer
- Candidate hosts
- 48, screened as one batch
- Screen
- Flux feasibility, Tier 2 target
- Advanced to design
- 11
- Reached construct
- 6
- Guide sweeps
- Cas-OFFinder, full host genomes
- Handoff
- Versioned export plus validation report
Forty-eight hosts to eleven on flux feasibility alone, at no compute cost. That first cut is where most of the value sits: thirty-seven designs that would each have consumed a greenhouse slot were eliminated on a budget argument the host’s own metabolic model made for them.
The eleven that advanced got real design attention — variant work on the controlling branch, guide design with off-target sweeps, construct assembly — and six produced packages worth building. The ratio is the point. Software narrows; the greenhouse decides.
04
Getting started
The 1,000-construct catalog across 65 organisms is browsable immediately and is the fastest orientation available — 14.1 Mbp of designed DNA with the reasoning attached. CPU engines are free, so an initial feasibility screen costs nothing but the time to set it up.
Teams running continuously generally move to an enterprise workspace for API keys, webhooks that push export packages straight into an internal pipeline, reserved GPU capacity so large sweeps skip the shared queue, and SSO.
Ready to run it
Try this in the Splice Platform.
CPU engines are free to run. GPU engines run on prepaid credits.
Related