Skip to content
rnaprog
This methodology article is published in English; the interface and SEO metadata are localised. Read the original English version.

Why synthesis vendors reject your sequence — and how to design around it

GC extremes, homopolymer runs, direct repeats and secondary structure are the four reasons an order gets bounced. Each has a design-time fix that costs less than a re-order.

rnaprog editorial 7 min read

A rejected order costs a week. Almost every rejection is one of four things, and all four are visible before you click order.

1. GC outside the vendor window

Most vendors quote a working range of roughly 25–75% GC for gene synthesis, and 40–60% for oligos. Sequences outside that range synthesise poorly: high GC produces strong secondary structure and hard-to-denature duplexes, low GC produces weak hybridisation and error-prone assembly.

The design fix is to set your GC window at design time rather than checking afterwards. If you codon-optimise without a GC constraint, you will drift — some optimisers push GC to 65% because the most frequent codons in many hosts are GC-rich. Constrain the optimiser to 40–60% and accept a slightly lower CAI.

For short oligos (siRNA, sgRNA spacers, ASO gapmers) GC matters even more: a 21-mer at 30% GC has a Tm near 50 °C and may not form a stable duplex at your assay temperature; one at 70% is prone to self-structure.

2. Homopolymer runs

Six or more identical bases in a row is the single most common flag. It causes polymerase slippage during synthesis and errors during assembly. Vendors differ on the threshold — many flag at 5, refuse at 8.

Design-time handling: after optimisation, scan for runs and substitute a synonymous codon. This is cheap for coding sequence (there are usually two or three alternatives) and impossible for a fixed target region — which is exactly why siRNA and sgRNA candidates get filtered rather than fixed. If your target region genuinely contains a 7 nt poly(A) tract, the honest answer is that this candidate is not orderable; take the next one.

Poly(T) deserves special mention for sgRNA: four or more T is a U6 termination signal, so a spacer containing poly(T) will truncate in the cell regardless of what the vendor does with it.

3. Direct repeats

Repeated stretches longer than ~20 nt cause recombination during assembly and are often rejected outright. Codon optimisation can create them accidentally — if the same amino acid run appears twice and the optimiser picks the same codons both times, you get an identical nucleotide stretch.

The fix is diversity-aware optimisation: when choosing codons for a repeated peptide, deliberately rotate among synonymous options so the nucleotide sequences diverge. This is one reason a candidate pool is useful — a run of twelve candidates will usually contain one without the repeat.

4. Secondary structure at the ends

Strong hairpins near the 5′ or 3′ terminus block assembly primers and reduce yield. Vendors rarely reject for this alone but will warn, and the synthesis may fail.

For mRNA constructs, structure near the 5′ end has a second cost: it impedes ribosome loading. A 5′ UTR chosen for low structure is worth more than a marginally better codon adaptation score.

Our folding estimate is a greedy base-pair model. It flags strong stems and hairpins reliably enough to be useful for ranking candidates against each other; it is not a thermodynamic minimum-free-energy calculation and should not be quoted as one.

A pre-order checklist

Before exporting an order sheet, check:

  1. GC within the vendor window — 40–60% for constructs, tighter for oligos.
  2. No homopolymer ≥ 6 — and no poly(T) ≥ 4 in anything destined for a U6 promoter.
  3. No direct repeat ≥ 20 nt.
  4. No strong stem within 30 nt of either end.
  5. Length matches what you intended — a truncated ORF is the most expensive kind of quiet error.
  6. Vendor-specific constraints — some refuse certain restriction sites, some require a minimum length.

All six are computable before you pay. The reason they still cause rejections is that people optimise for one metric and order immediately. The candidate table exists so you can look at the other five columns.

أول تشغيل يبدأ بـ 300 رصيد مجاني

يكفي لستة إنشاءات mRNA أو مجموعة من 12 siRNA، مع فحص الجودة والتصدير. دون بطاقة.

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