How to design a Gibson assembly that works

Updated October 2026.

Gibson assembly failures usually trace back to the design, not the bench. The reaction is forgiving. What is not forgiving is a junction with 12 bp of homology, two junctions sharing the same terminator, or a vector that was never fully linearized. This guide covers the numbers that matter, the junctions to avoid, and how to get from fragments to an oligo order.

What is Gibson assembly, and when should you use it?

Gibson assembly joins two or more linear DNA fragments in a single isothermal reaction, using shared sequence at the ends of neighboring fragments to define the order. Three enzymes work together at 50 °C: an exonuclease chews back the 5' ends to expose single-stranded 3' overhangs, complementary overhangs anneal, a polymerase fills the gaps, and a ligase seals the nicks. The method was published by Daniel Gibson and colleagues at the J. Craig Venter Institute in 2009 and is sold as a one-tube master mix by several suppliers.

Reach for it when you want a scarless join and no dependence on restriction sites: inserting a gene into a vector, swapping a promoter, building a fusion protein with no linker scar, or stitching two or three PCR products together. It is also the fastest route when your parts are one-off sequences rather than a curated library of standard parts.

Should you use Gibson assembly or Golden Gate?

Use Gibson when the junctions are arbitrary and the parts are used once. Use Golden Gate for many constructs built from a reusable part library, or for more fragments than Gibson comfortably handles.

Gibson needs no enzyme recognition sites, so nothing has to be domesticated, but every junction needs its own long homology region, which means custom primers for each new combination. Golden Gate cuts with a Type IIS enzyme and ligates through 4 nt overhangs, so a part made once can be reused in any position sharing the same overhang pair, and good overhang sets assemble ten or more fragments at a time. The cost is removing internal recognition sites from every part first. Our guide to designing a Golden Gate assembly covers overhang selection in detail.

How long should each overlap be, and what melting temperature should it have?

Design 20 to 40 bp of perfect homology at each junction, with an overlap melting temperature of at least 48 °C. That is where commercial kit instructions and most lab protocols converge, and it is the best predictor of whether the reaction works.

A few refinements worth applying:

For simple two or three fragment joins, kit instructions allow 15 to 20 bp. Below roughly 15 bp, annealing becomes unreliable and assemblies fail at the bench.
Keep overlap GC content between about 40 and 60 percent. A 20 bp AT-rich overlap can sit below 48 °C even though its length looks fine, so check the Tm rather than the base count.
Longer is not better. Overlaps past about 40 bp push primers into lengths that cost more and carry more synthesis errors, with no gain in efficiency.
The homology must be exact. A mismatch mid-overlap is often tolerated, but one near the junction ends is not worth the risk when redesigning is free.

How many fragments can one reaction take, and how short can they be?

Two to four fragments is routine and gives the highest yield. Five or six is achievable with clean, gel-purified, accurately quantified fragments and a longer incubation. Beyond that, efficiency drops fast enough that a two-step assembly, building two intermediates and joining them, is usually quicker.

Fragment size matters as much as fragment count. Keep every piece above about 250 bp. The exonuclease chews back both ends of each fragment, and on a very short piece the two chew-backs can meet and destroy it before it anneals. If a part is genuinely short, a promoter or a tag, do not give it its own fragment: build it into a neighbor's primer tail, or order it inside a synthesized block. Equimolar fragments are the usual starting point, with the vector at a slight molar deficit to the inserts.

Which junctions fail, and how do you avoid repeats and hairpins?

The junctions that fail are the ones whose overlap appears more than once in the assembly, and the ones that fold. Before ordering, search each overlap against every fragment and the finished construct.

Repeats are the most common cause of scrambled products. Reuse the same terminator twice, or the same promoter in two positions, or assemble tandem gene copies, and two junctions can share near-identical sequence, so fragments join in the wrong order or drop out. The fix is to shift the junction out of the repeat: move the split point 50 to 100 bp into unique sequence, or use a different terminator variant at one position.

Secondary structure is the quieter failure. An overlap that forms a stable hairpin, or that runs very GC-rich, will not present a clean single-stranded end for annealing. Avoid long inverted repeats, runs of more than four or five identical bases, and GC content above about 65 percent. The remedy is the same: slide the junction a few bases until the overlap is unremarkable.

How do you linearize the vector and design primers with overlap tails?

Linearize the vector either by restriction digest at the insertion point or by inverse PCR around the backbone, then make sure no uncut circular plasmid survives. Uncut vector is the single biggest source of background colonies.

For a digest, pick an enzyme that cuts the backbone once and no insert, run it to completion, and gel purify the linear band away from the supercoiled form. For inverse PCR, treat the product with DpnI to destroy the methylated template, then gel purify. Either way, run a no-insert control: transform the linearized vector alone, and a lawn of colonies means the problem is your vector prep, not your design.

Primers for Gibson have two parts. The 3' end is the annealing region that binds your template, typically 18 to 25 nt at 60 to 65 °C, calculated on the annealing portion only and never on the whole oligo. The 5' end is a tail carrying homology to the neighboring fragment. Put the full overlap on one primer of the pair, or half on each of the two primers meeting at that junction, so the two amplicons share the complete overlap once assembled. Keep total oligo length at or under about 60 nt: standard desalted synthesis is cheapest there, and longer oligos carry more synthesis errors. Our guide to designing PCR primers for cloning goes deeper on annealing regions, Tm matching and specificity checks.

Why did my Gibson assembly fail?

Match the symptom to the cause before redesigning:

No colonies at all. Usually too little DNA, a missing fragment, or a dead master mix. Quantify every fragment, confirm each on a gel, and run the kit's positive control.
Many colonies, all empty vector. The vector was not fully linearized, or the template plasmid survived the PCR. Gel purify the linear band and add a DpnI treatment.
Correct size by colony PCR, wrong sequence at a junction. Synthesis errors in a long primer tail. Sequence across every junction, not just the insert.
Scrambled or truncated assemblies. Repeated sequence shared between junctions. Move the junctions into unique sequence.
One fragment never incorporated. Often a fragment under 250 bp, or an overlap Tm that is too low. Merge it into a neighbor's primer tail, or lengthen that overlap.
Works with two fragments, fails with five. Assemble in two stages instead of forcing one pot.

What should you check before ordering oligos?

Run this once, before you spend money:

Every junction has 20 to 40 bp of exact homology with a melting temperature of at least 48 °C.
No overlap sequence appears anywhere else in the assembly.
No overlap contains a strong hairpin, a long homopolymer run or more than about 65 percent GC.
Every fragment is at least 250 bp.
Each primer's annealing region sits at 60 to 65 °C, measured without the tail.
The vector is linearized at the intended position, with a no-insert control planned.
Coding junctions are in frame, with no premature stop codon introduced by the join.
The assembled sequence has been simulated end to end and reads the way you expect.

How do you design a Gibson assembly in TeselaGen Community (free)?

TeselaGen Community is a free edition that includes assembly design, so you can plan a Gibson assembly, get the primers and export an oligo order form at no cost.

Get your sequences in. Import or paste the vector and insert sequences into the DNA Sequences library. Each opens in the sequence editor with its plasmid map, features and cut sites.
Prepare the backbone. In the sequence editor, filter cut sites for single cutters at the insertion point, then simulate the digest to confirm the linearized band size before committing to an enzyme.
Create a design. From the Design library, start a blank design: list layout for one fixed construct, combinatorial layout for several part choices per position.
Lay out the fragments. Add one bin per piece of the assembly, label each with its SBOL glyph (promoter, CDS, terminator), and insert the sequence or part that belongs in each bin.
Set the method to Gibson. In the reaction inspector, pick Gibson, then open its assembly parameters. The shipped defaults are 26 bp of homology per junction, a 60 to 70 °C overlap Tm window and a 250 bp minimum fragment size, all editable per design.
Clear validation. The validation panel flags junction and digest problems. Resolve them before submitting.
Submit for assembly. TeselaGen computes the junctions, designs the primers with their overlap tails, and generates the assembly protocol: input sequences and parts, oligos, PCR reactions, assembly pieces and finished constructs.
Save and order. Save the assembled constructs to the sequence library and the oligos to the oligo library, then export the oligo list as CSV or send the sequences into the built-in Twist or IDT order forms.
Check the frame yourself. Open the assembled construct in the sequence editor and turn on the translation track. The platform does not verify that a coding junction stays in frame.

CoScientist, the AI assistant included in Community, can start this from a description, setting up the design or drafting primers for you to review before submitting.

The full feature list is on the free Community edition page, and the wider design tooling on the molecular biology toolkit page.