TCIConsultancy

Questions about the tools

What these seven tools do, where your sequence goes, and how far to trust what comes back. Each tool also carries its own How this works and Reading your results panels; this page covers what they have in common.

Where your sequence goes

Six of the seven run entirely in your browser; Cleave-It, Clone-It, Assemble-It, Pack-It, Plan-It and Recode-It do all their work in JavaScript on your own machine. Nothing you paste is uploaded, nothing is stored, and no account is needed. You can save any of those pages and run them with no network at all. Each page loads a small counter that records that the page was opened — not what you put in it.

Annotate DNA is the exception, because it has to be. Annotation runs BLAST against a curated feature library that lives on the server, so your sequence is posted there to be analysed. Results are deleted from the server after three hours and the pasted sequence is deleted as soon as the run finishes. Download anything you want to keep. If a sequence is too sensitive to leave your machine, use the other six tools on it and keep Annotate DNA for material you are comfortable sending.

None of the tools share data with each other except when you press one of the Carry it through the rest of the bench buttons, and that hand-off travels inside your browser rather than through the server.

Run barcodes and reproducibility

Tools that make a choice among many valid answers — Recode-It above all — stamp each run with a barcode such as TPA-376489457. The barcode is not a lookup key for anything stored on a server. It is the seed: together with your input sequence, the parameters and the engine version, it determines the output completely.

That means a run can be reproduced from the record rather than from a saved file. Paste the same sequence, load the run record, press the button again, and the page regenerates the identical sequence and checks it against the hashes in the record. If anything differs — a changed codon table, a different cutoff — the replay is reported as inexact rather than quietly producing something else.

Keep the run record alongside any sequence you recode. It is the only thing that ties a construct back to the exact parameters that produced it.

Engine versions

Where a tool reports an engine version, that number identifies the algorithm, while the barcode identifies the inputs. The two are deliberately kept apart.

Algorithms are frozen once released. When one changes, the new behaviour is added as a new version and the old code stays in the page permanently. A record made under version 1 asks for version 1 by number and replays byte-for-byte on a page whose default is version 5. That is why old run records keep working, and why a tidy-up of released code is never a tidy-up — reordering even one step would invalidate every barcode ever issued under that version.

If a record names a version newer than the page you are using, the page says so instead of guessing.

Citing a run

For a notebook entry, a batch record or a methods section, quote the tool, the engine version where one is shown, the barcode, and the parameters that matter. For example:

Recode-It (TCI Consultancy, tci-consultancy.com/tools), engine v1, run TPA-376489457; Homo sapiens codon table, rare-codon cutoff 10%, CpG elimination on.

Downloaded files carry most of this already: FASTA headers include the barcode and the key measurements, GenBank output records the barcode and input fingerprint in its COMMENT lines, and the run record holds every parameter plus SHA-256 hashes of the input and output. Attaching the run record is better than quoting it.

What these tools are not

Every one of them is a design aid. None of them is a validation, and none of their output is a regulatory record.

For research use only. Verify critical designs before ordering synthesis, and before relying on any of this in regulatory, GMP or quality documentation.

Common questions

Why does a tool refuse my sequence?

The usual causes are a length that is not a multiple of three where a coding sequence is expected, an internal stop codon from the wrong reading frame, or ambiguity codes. Tools that need an ORF want the coding region only — start at ATG, end at the stop codon, no introns and no UTRs. A FASTA header line is fine and is ignored.

Can I paste a whole plasmid?

For Annotate DNA, Cleave-It, Clone-It and Pack-It, yes — that is what they expect. Pack-It finds the ITRs itself and measures between them, so there is no need to trim to ITR–cargo–ITR. Recode-It is the exception: it recodes one open reading frame and needs the coding region on its own.

Why do the sequences in a Recode-It panel look almost identical?

Because they have to. Every member encodes the same protein, so only synonymous positions can differ, most of those differ by a single wobble base, and methionine and tryptophan cannot differ at all. Two maximally distant members of a panel still align at roughly 80%. Judge the spread by the GC, CAI and changed columns rather than by an alignment.

Why does the same tool give a different answer on the same sequence?

Where many answers are equally valid, the tool samples among them, and the barcode records which sample you got. Paste the barcode back in to get the same one again. If the answer changes with the same barcode and the same parameters, something else changed — usually a pasted codon table.

Where do the built-in codon tables come from?

They are rounded reference values for Homo sapiens and E. coli K-12, included so the tools work out of the box. For anything you intend to cite, paste the exact table you want to use. The run record stores a hash of the table, so a replay against the wrong table is detected.

My browser freezes for a second on a long sequence.

Expected on multi-kilobase input. The heavy steps run on the page itself rather than on a server, which is what keeps your sequence local; the trade is that a long ORF ties up the tab briefly. Nothing is lost — wait for it.

Can I use these commercially, or in a regulated environment?

They are free to use for any purpose, with no registration. They are not validated software and carry no warranty, so output must be verified before it supports a regulatory, GMP or quality decision. If you need that level of assurance, or want the design reviewed rather than the arithmetic checked, that is consultancy work — get in touch.

Something is wrong, or a feature is missing from the library.

Please say so. The curated feature library is maintained by hand and grows by correction. Send the sequence or the feature name, and the run barcode if there is one.

Need more than a free tool?

TCI Consultancy advises on vector design, plasmid and vector CMC, assay development and the documentation that goes with them. If one of these tools has raised a question you would rather not answer on your own, that is what the consultancy is for.

Get in touchOur services

This page is plain text and does nothing but display it.