Verify Assembly — Check an Assembled Plasmid Against Its Parts

Different question, different input. Verify Construct re-derives only the PCR'd insert and asks whether it turns up correctly somewhere in the sequence you got back, so it says nothing about the rest of the plasmid. This page re-derives the whole product of the entire recipe — vector and insert, or all the fragments — and diffs your whole claimed sequence against it, which is what catches a wrong vector, a wrong backbone element or a bad junction outside the insert. The price is a heavier input: you have to declare the method, the parts and the enzymes, not just the PCR that made the insert.

Six: restriction–ligation, Gibson, Golden Gate, LIC, SLIC and In-Fusion/CPEC. TOPO is deliberately not among them, because a TOPO product is fixed by the charged vector rather than by the parts you can hand over, so there is nothing to independently re-compute. A method it does not recognise is refused outright with an error rather than quietly assembled as something else — a checker that silently substitutes a chemistry would report fabricated differences against a construct that is perfectly fine.

Because some recipes genuinely do. A non-directional ligation puts the insert in both orientations and half the plate carries each; a vector cut more than twice offers more than one backbone fragment; and a Gibson junction whose fragments already share terminal sequence has two honest readings — one homology arm counted once, or a genuine tandem repeat present twice. Your claimed sequence is checked against all of them, and the verdict names the one it matched, so "verified, and it is the reverse-orientation one" is a result you can actually get instead of a bare failure.

Normally no. When you mark the product circular, the check takes the first 24 bases of your claimed sequence and looks for them inside the predicted product — doubled so wraparound works, and in both orientations — then rotates and if necessary reverse-complements the prediction to line up before diffing. If that anchor cannot be found in either orientation the comparison still runs, but the result says so plainly and warns that the differences below may be an artefact of the rotation rather than real.

No — give bare parts for Gibson and Golden Gate. Gibson arms are added by the assembly primers and merged during the reaction, so fragments that already carry them make the concatenation one arm too long per junction. The convention is the opposite for SLIC and In-Fusion, where the homology must already be on the fragments because it arrived on the PCR primer tails (In-Fusion also works the assembly order out from that homology, so the order you list is ignored). If Gibson fragments do share a long terminal block, the assembly detects it, reports both readings and the check tries both, naming which one your sequence matched.

It is a sequence-level identity check against a re-simulation of the recipe you declared, and nothing more. It does not check off-target amplification outside a supplied PCR template, the biological function of the resulting protein or construct, or host and strain compatibility. It also cannot be more right than the recipe you gave it: name the wrong enzyme, or the wrong dNTP for a LIC chew-back, and it will faithfully verify your sequence against the wrong molecule. The LIC dNTPs in particular have no default for exactly that reason — T4 polymerase's exonuclease stops at the first occurrence of that base, so a guess moves every junction.

For the three overlap-directed methods — LIC, SLIC and In-Fusion — yes, and they are refusals rather than silent truncation: at most 24 fragments, 60,000 bp in any one molecule, and 500,000 bp across the inputs. In-Fusion scans terminal homology over every ordered pair of fragments to work out the assembly order, so these are work limits rather than biological ones, and they match the limits the cloning simulator applies to the same chemistries. Gibson and Golden Gate assemble in the order you list and are not capped this way.

Yes, though not from this page. Through the REST API and the MCP server, vectorPcr, insertPcr and fragmentPcrs each take a template, a forward and a reverse primer (plus an optional mismatch allowance and a circular flag), and that part is re-derived by in-silico PCR before the assembly is run — which is the stronger check, because a PCR that gives no product, or more than one, is itself reported as a finding instead of being assumed to work. This page takes the parts as pasted sequences.