Drybench Plasmid.

What the app shows, how it works and how it was built, in the words of its own write-up, followed by the parameter defaults and the data sources with their licences.

What it does

Paste a sequence, a FASTA or GenBank record, open a GenBank or SnapGene .dna file (read-only), start from pUC19 or pBR322, or fetch a public record from NCBI when you ask, and the plasmid is drawn as a circular or linear map: the parts the annotation library finds on either strand and across the origin, each with the alignment's identity, coverage, mismatches and gaps; the sites of 223 commercially available restriction enzymes with dam and dcm methylation as you set it; and the sites of the shipped sequencing primers.

The digest view gives the fragments of any digest and a schematic gel, and the picker finds the single and double digests whose bands tell the correct construct from one with the insert missing or reversed. The numbers view gives the plasmid's own double-stranded weight and the ng, pmol, copies per µL and ligation amounts that follow from it. A feature you add or edit is kept on the plasmid in this browser and travels in the GenBank file you export.

How it works

Annotation seeds every library part on both strands with shared 12-base words, clusters the seeds by diagonal and aligns each cluster locally with affine gaps in a band around it (Smith and Waterman with Gotoh's gaps; match +2, mismatch −3, gap open −5, gap extend −2). A match needs 99 percent identity and 95 percent coverage; a partial match 90 and 50 percent with at least 30 aligned bases; a part under 30 bases is called only when found whole and exact. A circle is searched extended by the longest part less one base, so a part across the origin is found once.

Restriction sites are found from each enzyme's IUPAC site on both strands and across the origin, with the cuts on each strand, Type IIS cuts outside the site and cuts that fall off the end of a linear molecule. Each site's dam and dcm marks are read against McClelland, Nelson and Raschke 1994, Table I: a site with no mark cuts; a mark on a base the table lists as not cut blocks it; marks on bases listed as cut leave it cutting; anything else reads "may be blocked" and is never counted as a cut. Everything runs in a Web Worker in your browser; the data files load from this site only once a sequence is open, and nothing is fetched from another host unless you ask for an NCBI record.

How it is checked

Every enzyme row was typed by hand and checked against Biopython's Bio.Restriction (an offline oracle) on pUC19 and pBR322; the 40 golden-set enzymes' pages were read by hand. The annotation engine agrees with Biopython's PairwiseAligner, with the same scoring, on 25 alignments of library parts to other NCBI records, within 0.1 percentage points of identity and coverage and with the same place, strand, call and score; synthetic inserts on each side of every threshold are called as the thresholds say, and random sequence and the golden plasmids reversed give no match. The bench numbers equal exact rational arithmetic, and the read-only .dna reader reads synthetic files built from NCBI records exactly as Biopython's reader does. The fixture files and their hashes are on the verification page.

Equations and defaults

Feature calls

match +2, mismatch −3, gap open −5, gap extend −2 (a gap of k bases scores −5 − 2(k − 1)). identity = identical columns / alignment columns (gaps included); coverage = the feature's aligned bases / its length. This app’s annotation-calling heuristics; match: identity ≥ 99 percent and coverage ≥ 95 percent; partial: identity ≥ 90 percent and coverage ≥ 50 percent, at least 30 aligned bases; under 30 bases only whole and exact. Seeds are shared 12-base words on either strand, clustered by diagonal; each cluster is aligned in a band of 16 bases plus 2 percent of the part's length. A part with no shared 12-base word is not seeded, which is why no call does not mean a part is absent. A partial match is drawn over its aligned bases only.

Badge words

Calling thresholds are This app’s annotation heuristics, not biological standards. Exact match at 100 percent identity and coverage; otherwise Match or Partial match with the identity, coverage, mismatches and gaps (gapped columns); a partial match adds truncated (under 95 percent of the library entry aligns), mutated (under 99 percent identical) or both. Identity and coverage are shown to one decimal, truncated toward zero, so a value shown never reaches a threshold the true value has not.

Restriction sites

Each enzyme's IUPAC site is searched on both strands, across the origin of a circle by extending the sequence by the site's length less one; each site is cut on the top and bottom strands at the row's stated offsets, so Type IIS enzymes cut outside their sites and a cut beyond the end of a linear molecule is reported as off the end.

dam and dcm

dam: G-m6A-TC on both strands; dcm: C-m5C-WGG on both strands; outcomes from McClelland et al. 1994, Table I, per base and strand. A site with no mark cuts; a mark on a base McClelland et al. (1994, Table I) record as not cut blocks it; marks only on bases recorded as cut leave it cutting; anything else, including slow, partial or conflicting entries and enzymes not in Table I, reads may be blocked and is never counted as a cut. Blocked by either methylation beats may be blocked, which beats cuts. This model calls DpnI cutting for fully methylated GATC only. DpnI can cleave hemimethylated sites much more slowly (NEB reports approximately 60-fold); that behaviour is not modelled.

Counting and cutter lists

A site count reads {n} sites: {c} cut, {b} blocked, {u} may be blocked. An enzyme with any site that may be blocked is listed apart, as depends on methylation, and is left out of the unique, two-cutter and no-cutter lists and of the picker.

The digest picker

The plasmid as given, with the marked insert removed (empty) and with it reversed: single digests of every enzyme that cuts the plasmid 1 to 6 times with no site that may be blocked, and pairs of them with at most 8 cuts together, are compared by their schematic bands. Two constructs are told apart when their band counts differ or a corresponding band differs by more than 5 percent; digests that tell all three apart come first, single enzymes before pairs, the clearest separation first.

The schematic gel

schematic: band order and log10(size) spacing; under 100 bp not drawn; sizes largest first, a fragment within 5 percent (inclusive, a judgement) of a band's largest fragment joins that band and may not resolve. A digest with one site that may be blocked draws two lanes, if it cuts and if blocked; with more than one such site no lane is drawn.

Primer sites

Exact and one-mismatch sites of the shipped primers on both strands and across the origin, sorted by position; a mismatch's place is counted from the primer's 3′ end (1 is the 3′-terminal base). The Tm is the primer's against a perfect match, from the sequence module (SantaLucia 1998 nearest-neighbour parameters, primer in excess at 250 nM, 50 mM Na⁺, no Mg²⁺), and is not given for a mismatched site.

Bench numbers

MW: both strands, residues as dNMP less water, CIAAW 2021 atomic weights; circular: no end groups; linear: 5′-OH ends. pmol = ng × 1000 / MW; copies = ng × 10⁻⁹ × N_A / MW (N_A = 6.02214076 × 10²³); insert ng = vector ng × (insert bp / vector bp) × ratio. The 650 Da per base pair rule is shown beside the exact weight as the approximation it is.

Inputs

Pasted letters, FASTA and GenBank records (the first record of a file), and SnapGene .dna files read-only by a reader of the public packet structure that knows the cookie, DNA, features, primers and notes packets and refuses any other, a truncated file, a declared length that disagrees with the sequence and an unknown format version. Up to 50,000 bp; IUPAC letters other than A, C, G and T are kept and never match a site.

Data sources, licences and versions

Times on screen are local time with the UTC offset; times in exports are UTC. Dates with no time are shown at the precision the source gave them.

The Export menu offers the features, the enzyme sites, the digest's fragments, the primer sites and the numbers as CSV with the provenance sidecar; the map and the schematic gel as SVG or PNG with the provenance embedded; a printed page with the map, the feature table and the digest; the plasmid as a GenBank file with every feature, and as FASTA. A GenBank file written here is never presented as an NCBI record: its ACCESSION and VERSION lines are a single full stop, the source record is named in its COMMENT, each feature the library found carries /provenance="library" with the library version, the badge and the licence in its /note, and a feature you added or edited carries /provenance="added" or "edited" and none of the source's identifiers.

CSV columns

Features

name, type, start, end, strand, length_bp; call (Exact match, Match, Partial match or empty for a feature that is not a library call); identity_pct and coverage_pct to four decimals, truncated; identical, columns, mismatches, gaps (gapped columns), gloss and aligned_bases; local_alignment_score, the local alignment's score under the scheme in the methods (match +2, mismatch −3, gap open −5, gap extend −2; a gap of k bases scores −5 − 2(k − 1)), for re-deriving a call and never a quality measure: for a feature under 30 bases no alignment runs and it is set to twice the length; provenance (library, record, file, added or edited); source_accession, source_location, licence; and uniprot, the accession only when the call carries it (a full-length match whose translation is identical to the entry's canonical sequence).

Enzyme sites

enzyme, site, position (the site's first base), strand, cut_after_top, cut_after_bottom; cut_call (cuts, blocked or unknown); methylation (none, blocked, not-blocked, not-recorded or needs-methylation); methylase (dam, dcm or both); host_methylation, one of the tokens dam-dcm (E. coli, dam positive and dcm positive), dam (dam positive, dcm negative), dcm (dam negative, dcm positive), none (dam negative, dcm negative) or unmethylated (a PCR product or synthetic DNA); methylation_source.

Fragments

size_bp, from, to, left_enzyme, right_enzyme, and variant (if it cuts or if blocked, where one site may be blocked); the host is in the provenance.

Primer sites

primer, position, end, strand, mismatches, mismatch_from_3prime (1 is the 3′-terminal base; empty at an exact site), tm_c (empty at a mismatched site).

Numbers

quantity, value, unit and formula, each row as shown.

Release gate: before the app leaves the preview, its exports and printed reports will carry the data versions, parameters, limits and a timestamp needed to reproduce a result, as the regulatory position says. This section will describe each format as it ships.

A link carries a public example, an NCBI accession or the sequence of a plasmid up to about 5,500 bp, with the chosen enzymes, the methylation setting, the data versions and the view; it opens identically next year because it names the versions it was made with. Use links for public examples, accessions and unedited plasmids of that size.

A GenBank file carries everything else: a larger plasmid, the features of a file you opened, and your own feature edits, each marked as yours. Copy link says so when the view holds something a link cannot carry, and offers the GenBank file first. On the same browser, Send to carries a plasmid of any size to another app.

Each module below is a set of automated tests in the app's source code, with the expected results written into the tests. They check the app's logic against recorded inputs and stored reference records; none of them calls a live service.

src/lib/seq/digest.test.ts

Every enzyme row against Biopython 1.88 Bio.Restriction (its data are REBASE release 404, an offline oracle only): 223 enzymes on pUC19 and pBR322 (446 cases), cut positions exact, the 40 golden-set counts against the enzymes' pages read by hand, 6 double digests, the sites across the origin, Type IIS cuts off the end of a line, and the picker's pool and verdicts on the correct, empty and reversed trio.

src/lib/seq/methylation.test.ts

19 cases worked by hand from McClelland et al. 1994, Table I, and the cited blocked examples on pUC19 and pBR322; the reasons by category; blocked beats may be blocked.

src/lib/seq/features/annotate.test.ts

26 alignments against Biopython's PairwiseAligner (identity and coverage within 0.1 percentage points; place, strand, call and score exact), 12 synthetic boundary cases, and the specificity test: no match and 0 partial calls on 10 seeded random sequences of 10,000 bp and both golden plasmids reversed.

src/lib/seq/features/primers.test.ts and snapgene.test.ts

Each shipped primer at its NCBI binding site; one-mismatch places on both strands and across the origin. The .dna reader against Biopython's SnapGene reader on 6 synthetic files built from NCBI records, every feature recovered, and 4 refusal files refused for their stated reasons. Real SnapGene files are checked only at the pre-live pass.

src/lib/seq/dna-amounts.test.ts

11 cases against exact rational arithmetic, to full double precision.

src/components/tools/plasmid-map/genbank-export.test.ts

The GenBank export read back by Biopython 1.88: ACCESSION '.', /provenance kept, no identifier on a feature you added or edited, every /note ASCII, and every CDS's /translation equal to Biopython's or left out where Biopython refuses it.

src/lib/seq/plasmid-examples.test.ts

The worked examples' golden fixtures, written once by the oracle: every call, the UniProt check of each linked CDS call, every primer site and every enzyme's cut count on pUC19 and pBR322.

pipeline/plasmid-map/reference.py and make-dna-fixtures.py

The oracle script (Python 3.11, Biopython 1.88) and the build-time .dna writer, MIT; the fixtures' hashes are in fixtures/manifest.json, and the writer is never imported by the app.

Scientific computation runs in your browser, and individual apps need no account. Optional Bench accounts and backups use the site server and configured providers. The privacy statement, the hosts it contacts and the headers its workspace is served with are generated from its registry entry, the same record that sets the browser's policy.

What leaves your browser

Hosts the app contacts

• eutils.ncbi.nlm.nih.gov

  Operator

  NCBI, US National Library of Medicine

  What for

  Fetching a public nucleotide record by accession, only when you press Fetch from NCBI.

  What it receives

  The accession you typed or a link armed, and your IP address

  When

  only when you confirm

  • one accession: 1 request

Headers the workspace is served with