Skip to content
rnaprog

À propos de rnaprog : pourquoi un outil et pas un chat de plus

Beaucoup d'outils s'arrêtent à la table des codons. Nous calculons toute la chaîne (optimisation, UTR, structure, QC) et consignons les paramètres pour reproduire.

What this product is for

Most RNA projects lose their first month before any bench work starts. Someone has to take a protein or a transcript, pick a codon table, hand-build a few constructs in a spreadsheet, check GC and repeats by eye, paste sequences into a vendor order form, and then discover the vendor rejects one of them for a restriction site nobody noticed.

rnaprog compresses that month into a run: give it a target and a constraint set, get back 96 ranked candidates with CAI, GC, a folding estimate, region annotations and a vendor-ready export. An mRNA construct design costs 40 credits; a 12-candidate siRNA pool costs 36. The point is not that the tool knows which molecule will work — it is that the first sixty candidates stop being a spreadsheet exercise.

It is deliberately narrow. It does not predict expression, it does not search a whole genome for off-targets, and it will not tell you whether your construct is patentable or whether your institution will approve the work. Where the tool is the wrong instrument, we say so on the use-cases page rather than in a support thread.

Five operating principles

Scores rank candidates; they are not measurements

CAI, GC, ΔG and Tm are computed from published models and stated parameters. They tell you which candidate is more likely to behave, not how a molecule will behave in your cells. Every report says which model produced which number.

Every candidate traces back to its inputs

A run stores the codon table version, the seed, the GC window and the constraints that produced each sequence. Six months later you can re-derive the same pool instead of guessing why candidate #4 looked different.

Failures cost one run, not one project

A run that dies before producing candidates is refunded in credits automatically. Design work should not be lost to a queue timeout at candidate 90 of 96.

Say what the tool cannot do

No expression prediction, no whole-genome off-target search, no guarantee a construct works. Folding is a greedy heuristic, not an MFE calculation. Those limits are on the product pages, not in a footnote.

Biosecurity screening is not optional

At submission we run a rule- and keyword-based automatic initial screen; it does not replace the screening your synthesis vendor applies. A hit cancels the run, refunds it in full, and can be appealed at /appeal. A prohibited design ends the account. This is the one area where we do not optimise for conversion.

Compliance, in plain language

These are the four areas that actually stop a project. None of them are edge cases.

Synthesis vendor screening

IDT, Twist, GenScript and every other vendor apply their own biosecurity screening and may refuse an order. Our export formats match what they accept; a refusal is their decision, not a defect in the sequence.

Sequence provenance and freedom to operate

You must have the right to use any sequence, plasmid map or accession you upload. We do not run a patent or FTO search, and a designed candidate is not a defence against an infringement claim.

Institutional approval for controlled work

Work on controlled pathogens or dual-use research requires your institution’s approval, held by you. We may ask for the approval reference on the account before allowing such runs.

Data handling

Sequences and reports stay in your account and are never used for model training. Payment credentials never reach our servers — cards are handled by Creem and Waffo Pancake as merchants of record, crypto by on-chain verification only.

Full text lives in the terms of service, acceptable use policy and privacy policy.

Who runs it

A small independent team. We run our own designs through the same pipeline and publish the credit numbers from those runs. There is no sales team, and no enterprise tier that unlocks a better algorithm — every plan runs the same engine.

Design engine

Codon optimisation, region constraints, folding heuristic, off-target seed scan

Reports & export

QC panel, FASTA / GenBank / order-sheet output, 12-language interface

Payments & compliance

Three channels, sandbox/live switching, biosecurity screening

Votre première exécution commence avec 300 crédits offerts

De quoi faire 6 constructions d'ARNm ou un pool de 12 siRNA, QC et export inclus. Sans carte.

300 credits · No card required · You keep the rights to your sequences