Cloning workspace

Why keep the whole strategy in one file?

A construct is usually the product of five or six reactions, and the reason a clone cannot be reproduced two years later is almost never the final sequence: it is that nobody wrote down which vector was digested with what, which primers amplified the insert, and which of three candidate products was actually taken forward. A cloning strategy file records the graph rather than the endpoint. Talindrew saves OpenCloning's cloning-strategy JSON unchanged, so the record is an open, documented format that a second tool can already read, and the zip variant carries the sequencing traces that verify the product beside the design that produced it.

How is this different from the cloning wizards?

The wizards are still there and unchanged: pick Gibson or Golden Gate, walk the steps, load a worked example, get a product. They are the fastest route through one reaction. The workspace is for the other shape of problem, where the answer is a chain: amplify two fragments off a genome, digest a vector, assemble all three, then recombine the result into a destination plasmid, with every intermediate still on the canvas and still editable. Deleting a step cascades to everything downstream, and reopening a strategy puts every node back where it was.

• One reaction, guided, with a demo to copy: use a cloning wizard.
• A multi-step build you want to keep, share and reproduce: use the workspace.
• A strategy someone sent you as OpenCloning JSON: open it in the workspace.

What can each step do?

Every node is either a way of getting a sequence in or an operation on the sequences already on the canvas. The operations are the sixteen OpenCloning defines, run by the same engine.

• In: manual entry; a GenBank, FASTA, EMBL or SnapGene file, including the cloning history embedded in a .dna; an id from Addgene, iGEM, SEVA, Euroscarf, WeKwikGene, Benchling, SnapGene or Open DNA Collections; a genome region from NCBI by assembly, gene or coordinates.
• Single-input operations: restriction digest with a product chooser that shows the fragment on the parent and spells out its sticky ends, PCR, oligonucleotide hybridisation, polymerase extension, reverse complement and annotation.
• Assemblies: Gibson, In-Fusion, In vivo and overlap-extension; restriction and ligation, which is Golden Gate; and plain ligation, blunt or sticky.
• Recombination: homologous recombination, CRISPR-HDR with guides taken from your primers, Gateway BP and LR, Cre/Lox, and recombinases with custom site definitions.
• Where a reaction has several valid products, they are offered side by side with a mini-map each, and the one you choose is what the graph carries forward.

Designing the primers inside the strategy

A PCR step can send you straight into a primer designer that knows what the primers are for. Ask for a normal pair and you get one; ask for restriction and ligation and the tails carry the enzyme's recognition site with the filler bases it needs to cut near an end; ask for a Gibson-family assembly and the homology arms come from the sequences being joined; ask for homologous recombination or CRISPR and the arms are built around the insertion point; ask for Gateway BP and the attB tails come from the known pDONR combinations. The designer previews the product it will make before you commit, and saving writes the primers back onto the PCR step they were designed for. In the Primers tab every primer carries primer3 numbers: Tm, GC, homodimer and hairpin, plus a heterodimer figure for each PCR it takes part in, with warnings on GC outside 30–70 %, Tm outside 50–70 °C, or a ΔG below −8000.

Verifying the clone from sequencing traces

Attach AB1, SCF or ZTR files to any sequence in the strategy and they are aligned against it on the server, best strand per trace, including reads that run across a plasmid's origin. What comes back is the aligned span drawn as a coverage track on the node's own map, in both the circular and the linear view, green where the read matches and orange where it does not, with percentage identity, a table of the mismatched positions and the chromatogram itself. A run of missing bases longer than 10 bp counts as uncovered rather than as deletions, which is the difference between a clean read and a thousand phantom mismatches. In this deployment the alignment is computed with Biopython rather than the MAFFT and MARS binaries the upstream engine shells out to; the stored result keeps the OpenCloning alignment shape either way, and the traces travel inside the exported zip.

Golden Gate kits, syntaxes and combinatorial assembly

Kit templates load from the OpenCloning submission repository, MoClo YTK among them, as a strategy with placeholder parts: each slot names a category, and picking a plasmid for it runs that part's own source. The Assembler tab goes the other way round: load a syntax, or upload one, and it walks the category chain filtering each step by the previous part's right overhang, takes a multi-select of plasmids per category, and assembles every valid combination at once as circular restriction-and-ligation products. Your own part plasmids can be uploaded and classified against the syntax before they are used, and the results table downloads as a zip of csv/tsv, GenBank and JSON. A syntax builder sits beside it for defining parts, overhangs and the overhang graph, validating that the graph is connected and forms a single cycle, and saving it as JSON.