Files and storage

Formats

The format is sniffed from the content first, so a GenBank record in a .txt file still opens as GenBank. If a file holds several records, the first one is opened, and a warning in the status bar says how many others there were. AB1 and FASTQ files are sequencing reads, with a quality for every base: see Sequencing reads.

A WGS, TSA or TLS master record (a CONTIG line, no ORIGIN, a LOCUS length counted in rc) holds no sequence of its own. It opens as an empty document, with a warning, and saving it writes a valid header only; the CONTIG line is not kept.

GenBank is the native format: sequence, topology, definition, accession, references, and every feature with its full location (complement(...), join(...), order(...), partial ends, ranges that wrap the origin) and qualifiers. An order(...) stays one, through edits and back out to a file; the one location PlasmidPop cannot keep is one operator nested inside the other, such as a join(...) inside an order(...), which is read as the outer one with a warning. A download writes a standard GenBank flat file that other programs can read. Anything PlasmidPop does not understand in a record is reported as a warning in the status bar rather than silently dropped. So is a record that contradicts itself: a CDS whose stored /translation is not the protein its own bases give (see Checking a record against itself). A qualifier whose quotes do not pair up — /note="One missing ""quotation mark" here", written by hand or by another program — is closed at the next feature, or at the next qualifier when its line ends in a quote, with a warning naming where; the features after it are read as usual. A REFERENCE line that gives no number or no location in parentheses (REFERENCE .) is written back as a plain numbered reference, and loose text on it goes to the reference's REMARK with a warning, so the file written is one other programs (Biopython among them) read.

A molecule with sticky ends — a fragment from a digest, say — has something GenBank cannot express, so the ends travel as a PlasmidPop-ends: comment line. Other software sees an ordinary comment; PlasmidPop reads it back as the document's ends (see Simulated cloning). A FASTA file carries the same text in brackets at the end of the header line (>insert [PlasmidPop-ends: left=5' AATT/EcoRI; right=blunt]), beside the usual [topology=circular]. A SnapGene file's own record of its overhangs is read on import — a linearised TA vector opens with its 3′ T at each end, a D-TOPO vector with its 5′ overhang.

Where the DNA was grown travels the same way, as PlasmidPop-methylation: dam-; dcm+, and only when it is not an ordinary dam+ dcm+ plasmid (see Where the DNA was grown).

What a product of a simulated reaction was made from travels as a block of PlasmidPop-made-from: comment lines, one per molecule of its tree (see What a product was made from).

A line of ours that cannot be read — edited by hand into something that no longer parses, say — is kept as an ordinary comment rather than dropped, in GenBank and in a FASTA header alike, so it is still there to fix.

ApE files are GenBank with extra colour qualifiers; those colours are used for the features.

FASTA gives a bare sequence. A header ending in [topology=circular] makes the sequence circular. Sticky ends and where the DNA was grown travel in the header too, as [PlasmidPop-ends: ...] and [PlasmidPop-methylation: ...], the same text as the GenBank comment lines. What a product was made from does not: it is too long for a header line, so a FASTA export leaves it out. Gaps and digits are stripped with a warning. A record whose letters are amino acids opens as a protein; see Proteins.

SnapGene .dna files are read for sequence and topology, features (including segmented features and their colours), primers (as primer_bind features), sticky ends, notes (description, organism, references) and the file's own history: how SnapGene recorded the construct being made shows in Made from, with the oligos of a PCR and a mutagenesis, and SnapGene's own word for anything PlasmidPop does not run (flip, newFileFromSelection, changeMethylation). SnapGene stores no checksums, so each molecule in that tree reads as _not in this browser_ unless you have the file yourself. Enzyme sets, alignments and appearance settings are skipped. A .dna whose double-stranded flag is clear opens as ss-DNA, otherwise as DNA. File > Download SnapGene… writes a .dna back (see Downloading a SnapGene file). Files from current SnapGene versions are what is tested; before version 1.2, primers were placed one base to the left of where SnapGene has them, and a primer SnapGene stores twice came in twice — reopen the .dna file to get them right. A feature or primer whose position cannot be placed on the sequence (outside it, or backwards on a linear one) is left out, and a warning in the status bar names it; the rest of the file opens as usual.

SnapGene .prot files open as protein documents: the residues, every feature (domains, regions, sites, with their colours) on the residues SnapGene shows them on, and the notes, as a .dna file's are. A protein downloads as GenPept, or as a .prot.

SnapGene .rna files open as a nucleotide document with the molecule type RNA: the bases, every feature (rRNA, tRNA, modified bases, variations) and the notes, as a .dna file's are. SnapGene writes an RNA with T where the RNA has U, and so does the editor underneath, as GenBank does: the stored sequence, every download and the clipboard keep T, and the molecule type RNA (named on a GenBank download's LOCUS line) says it is RNA. The sequence view draws it with U (see Viewing). The file does not say whether the RNA is single- or double-stranded, so nothing here does either, and it opens linear.

Geneious files are not supported. Export them as GenBank first.

Opening several files, or a ZIP

File ▸ Open takes several files at once (select them in the picker), and so does dropping them on the window. Each sequence opens in a tab of its own, and one summary says how many documents opened, which files were skipped and why, and any warnings the reader had about each record.

A .zip is opened the same way: a lab moving off Benchling, or another tool, can export all its sequences as a ZIP of GenBank files and open that one file. Every sequence file inside becomes a document. Files that are not sequences (a spreadsheet of metadata, say), empty or unreadable files, and the hidden files macOS and Windows add to archives are left out; the ones that could have been a sequence are listed under Skipped with the reason. A file with several records (a multi-FASTA) opens every record; an AB1 or FASTQ opens its first record only, as it does on its own.

The folder a file sat in is kept: it is added to the document's keywords (GenBank KEYWORDS) as Projects/Plasmids. Files with the same name from different folders are different documents and stay apart; a file that is identical to one already open, or to another in the batch, is not opened a second time and is counted as already open.

Opening hundreds of files gives hundreds of tabs, and the archive is read in the browser: nothing is uploaded. An archive that holds more than 5,000 files, or that would expand to more than 512 MB, is refused rather than opened, and a single file over 64 MB inside one is skipped. Detect features on open (if you have turned it on) does not run for a batch.

Opening a record from NCBI

File ▸ Open from NCBI… (or From NCBI… in the toolbar when nothing is open) fetches a GenBank record by its accession number, so a published plasmid, gene or protein opens without downloading it first.

1. Type the accession: L09137 (pUC19), NC_001422 (phiX174), NM_000581.2, or a protein's, NP_000509 (haemoglobin β). The version after the dot is optional; without it you get the current one. A pasted NCBI address such as https://www.ncbi.nlm.nih.gov/nuccore/L09137.2 works too.
2. Press Open (or Enter). The record opens in a tab of its own, as if you had downloaded it from NCBI and opened the file: every feature, the references and the comments. The tab takes the name on the record's LOCUS line, as it would from the file — L09137 opens as SYNPUC19CV — and Recent files lists it as from L09137.gb. Click the name in the toolbar to rename it. Opening the same accession again goes to its tab.

• Several at once: separate them with spaces or commas (up to 20). Each opens in its own tab, the last one in front. One NCBI has no record for is named in the status bar; the others still open.
• Cancel stops a fetch under way; press it again, or Escape, to close the dialog.
• Proteins (NP_000509, XP_…, WP_…, or a GenBank protein id such as AAA12345) come from NCBI's protein database as GenPept and open as protein documents, their tab listed in Recent files as from NP_000509.gp. Nucleotide and protein accessions can be mixed in one list: PlasmidPop then sends two requests, one to each database, and opens the nucleotide records first. If one of the two fails, the records the other brought still open, and the status bar says which accessions did not and why.
• Anything that does not look like an accession is refused before anything is sent, so a typo costs no request.
• Records up to 10 Mb. A longer one — a bacterial chromosome's scaffold, a whole contig set — is turned away as soon as its first line arrives, since PlasmidPop is not a genome browser.
• NCBI takes three requests a second from any one address without an API key. PlasmidPop sends all the accessions typed in one request (two when they mix nucleotide and protein, spaced apart) and tries once more after two seconds if NCBI is busy; if it still is, the dialog says so. It needs a connection: offline, it says that instead.

What is sent. This is the one feature that reaches a server other than the one the app is loaded from, and only when you press Open. What goes to NCBI (eutils.ncbi.nlm.nih.gov) is the accession numbers you typed and the name PlasmidPop, which NCBI asks every program using its service to send. No document, sequence or file name goes, no cookies, and not the address of the page you are on; there is no API key and no e-mail address. NCBI sees the request come from your network, as it would if you opened the record on its website. The usage statistics count that a record was opened from NCBI, never which (see Usage statistics).

Several documents at once

Every document you open gets a tab in the strip under the toolbar, so a vector and its insert can be open side by side. Open file, New, Open example, Open from NCBI…, a dropped file (or a ZIP of them), a pasted record, a ligation product and a fragment opened from the Cloning tab all open in a new tab and bring it to the front. Opening a file that already has a tab goes to that tab instead of opening it twice — unless that tab has been edited, which makes it a working copy rather than the file — and a "New" document nothing has been typed into gives its tab up to the next file you open.

• Click a tab to switch to it, or press Alt+1 … Alt+9 for the first to the ninth. Each tab keeps its own selection, undo history, enzyme ticks, cut sites, find bar and sidebar tab, and comes back scrolled to the row and with the map zoomed as you left it; the view switcher, the toggles and the Format and Edits settings are the same for all of them.
• Drag a tab along the strip to move it, or press Alt+Shift+PageUp / Alt+Shift+PageDown to move the one in front. Alt+1 … Alt+9 follow the new order.
• The × on a tab, Alt+W or File ▸ Close closes it. The document stays in the browser's storage and under Recent files, with its undo history: reopening it brings both back (see History). (Ctrl+W is the browser's, and closes PlasmidPop itself.)
• The Files tab at the left is the start screen with the recent files; the logo and File ▸ Show files go there too. The other tabs stay open behind it, and entries that are open in a tab are marked open.
• + at the right of the strip starts a new sequence, like New.
• A dot after a tab's name means that document has changed since it was last downloaded. Closing the tab, or the page, loses nothing: the document is in this browser's storage either way.

The open tabs, their order and the one in front are remembered, so a reload brings them all back.

Working copies

The file you open is never written to. The first edit to a document that came from a file turns that document into a _working copy_: it takes a name of its own — pBR322 becomes pBR322 copy — and a line under the toolbar says which file it came from and what this copy is called. Whatever happens in that tab afterwards, the file on disk is exactly as it was.

• Rename, on that line, gives the copy a name of your own; so does clicking the name in the toolbar. That name is what a download will be called, so it is worth setting early. If your first action is a rename, the copy is called what you called it and never carries a copy name at all.
• See what changed, at the right of the line, shows the differences against the original at any time.
• Opening the same file again gives you the original back in its own tab, so the two can be compared side by side.
• The copy's history starts at the contents of the file, under the copy's name. Undo therefore takes you back to what the file holds and stops there — there is no state in which the tab is the original again, and no edit that renames the document behind your back.
• A document from New, a paste, the bundled example, a ligation or a fragment opened from the Cloning tab has no file on your disk to protect and is never forked.

What a copy remembers about its original

A working copy also writes down what it came from, and carries that into every file and link it leaves as:

COMMENT PlasmidPop-derived-from: cdseguid=dUxN7YQ… pBR322.gb

The file name is a courtesy; the checksum is the part that holds up. A name is what somebody called a file once, while a checksum is the molecule itself and stays true however the original is rotated, renamed or exported again — so whoever ends up with pBR322 copy.gb can put the original beside it and _check_, rather than take the line's word for it.

• PlasmidPop says so under the toolbar when you open such a file: Derived from pBR322.gb, with the checksum. Point Compare with… at that file and the dialog confirms it is the one, then shows everything that has happened since.
• Other software sees an ordinary comment and ignores it. PlasmidPop takes the line out of the comments on reading and puts it back on writing, so a file never collects copies of it.
• It records the file a copy was forked from, not a chain: a copy of a copy names its immediate parent.

Downloading

PlasmidPop never writes to a file on your disk. Documents live in this browser as you work on them (see Local storage and recent files); when you want one as a file, you download it, and the file is yours from then on. Downloads are GenBank, or SnapGene on request.

• Download GenBank… (Ctrl+S, and Ctrl+Shift+S out of habit) writes the document in front.
• Download SnapGene… writes it as a SnapGene file instead; see below.
• A working copy is reviewed first: what it changed about the file it came from — a summary line, the whole molecule drawn as a map with the changes marked on the ring, each changed stretch of sequence drawn with the same marks the sequence view uses for tracked changes and headed with where it is and what happened there (around 1,204 inserted 5 bp; deleted 3 bp), and the features added, changed or removed. Each feature is named and says where it is (− misc_binding 411..414), a removal included; where one deletion took several features with it, the review says so once, naming them. A changed feature says what changed — ~ tet type gene → CDS — with qualifiers counted rather than quoted, since a /note can be a paragraph. Download writes it; Cancel writes nothing.
• In Chrome and Edge the browser then asks where to put the file, with the name filled in — the name you chose last time for this document, so you can point at the same file again and replace it. In Firefox and Safari a page may not ask, so the file lands in your downloads folder under a name the browser decides (see below).
• The name PlasmidPop offers comes from the document's own name, never from the file the copy was made from: pBR322 is offered as pBR322_copy.gb. Rename the document — click its name in the toolbar, or Rename on the working copy's line under it — to change what is offered, or type another name in the browser's dialog.

The dot after the document name means it has changed since the last download.

Downloading in Firefox and Safari

Chrome and Edge can ask a page where to put a file; Firefox and Safari cannot, so every download lands in your downloads folder with the browser naming it. Download the same document twice and you have pBR322_copy.gb and pBR322_copy(1).gb: two downloads of the same document at different moments, not two versions PlasmidPop made. A line under the toolbar says so the first time; Got it puts it away for good.

Firefox can be told to ask, and then you can replace the earlier file yourself:

1. Open about:preferences (☰ ▸ Settings), General ▸ Downloads.
2. Choose Always ask you where to save files.

Every download then opens the system dialog with the name filled in; point it at the file you downloaded before and confirm Replace. Safari has the same setting under Settings ▸ General ▸ File download location ▸ Ask for each download.

Nothing is lost in the meantime: every open document is written to this browser's own storage as you work and comes back when you return. Downloading is how you get a file _out_ of PlasmidPop, not how you avoid losing work.

Downloading a SnapGene file

File > Download SnapGene… writes a .dna file that SnapGene and SnapGene Viewer are meant to open: the sequence and topology, the Dam and Dcm methylation of the host, the overhangs of a linear fragment, every feature with its segments, colour, direction, reading frame and qualifiers, primers (a primer_bind that holds a primer's own sequence and notes goes into SnapGene's primer list; one with other qualifiers stays a feature), and the notes: name, description, accession, organism, comments, references and date. A protein goes out as .prot, and a linear RNA as .rna (with T for U, as SnapGene writes it). A circular RNA has no file of its own and goes out as a .dna. The double-stranded flag is set for DNA and left off for ss-DNA and for an RNA written as a .dna, so a single-stranded molecule re-opens as ss-DNA. A feature with no colour of its own gets the colour the map draws it in. source features stay in the file and in SnapGene's Features list but are marked hidden, as SnapGene does, so they do not draw as rings over the genes on its map; read back here they are the same features.

SnapGene's format has no place for everything a document can carry. Before the file is written a dialog lists what it will leave out or change, so nothing goes silently: GenBank-only header fields (keywords, version, lineage, database links), a molecule type SnapGene cannot keep (mRNA and ss-RNA come back as RNA, ds-DNA as DNA, a circular RNA as ss-DNA), an empty note on a primer, base colours and sizes, the bases and qualities of a sequencing read, partial-end marks (< and >), locations between two bases, order(...) locations (written as joined), the record of what a molecule was made from, and the enzymes behind a fragment's ends. If there is nothing to list, and the document is not a working copy, the file is written at once. A working copy still shows what it changed first, as for GenBank.

SnapGene's history tree is not written: the file records the molecule as it is, not how it was made. The file's format is not published, so it was written from SnapGene's own sample files and checked by reading the output back with PlasmidPop and with Biopython; it has not been opened in SnapGene itself. If SnapGene complains about a file, GenBank is the format to fall back on, and an issue with the file attached helps.

Verifying clones

File ▸ Verify clones… checks a plate of whole-plasmid sequencing consensuses against the open constructs they should be, and says which match; see Verifying a plate of clones.

Comparing with another document

File ▸ Compare with… (Alt+K) shows how the open document differs from another one: the same review a working copy gets before it is downloaded, asked of any file or tab rather than the one this document came from.

With other documents open it first asks what to compare with: one of the other tabs, as it is now, or A file on disk…, which opens the file picker. With no other tab open it goes straight to the picker.

It answers the question a plasmid map on its own cannot — _is this the same construct as the one in that file, and if not, where do they part company_ — which is what a colleague's copy, a vendor's sequence or last month's version of your own plasmid is usually for.

• Nothing is opened, written or stored unless you ask. A file is read, compared and dropped; no tab appears and neither file changes. The document you are looking at is the one that keeps its tab. To work on the other file too, Open _file name_ in the dialog opens it in a tab of its own, exactly as Open file… would; for another tab, Go to _tab name_ brings it forward.
• The differences are in this document's coordinates: what it has that the other does not. The map at the top marks them on the ring, so where they fall is the first thing you see, with a feature the other has and this one does not drawn as a broken red ghost where it would be. Each neighbourhood of changed bases is drawn the way the sequence view draws tracked changes, and the features added, changed and removed are named below them.
• The map points into the list. Click a mark or a wedge on it and the review scrolls to the stretch of sequence it is in and lights it for a moment; click a ghost and it goes to that feature's line. The other way round, a removed feature's name in the Features list is a link: click it and its ghost is picked out on the map. Neither changes your selection — the review is not the editor. The same holds in the review before a download.
• Features are matched by what they are, since two files give the same feature different internal ids. A feature both agree on to the last qualifier is not reported. One that differs is matched to the feature it became when it is still in the same place and still recognisable, by its name or by its type, and then it is one changed line rather than a removal and an addition at the same coordinates. Two features that share only a place are not matched: a gene and the CDS inside it cover the same bases and are not versions of one another. A feature that both moved and was renamed is matched by its bases instead: the same type on the same strand covering the same sequence, at least 20 bp of it, when that sequence is annotated once in each file. Two copies of a repeated element are left as a removal and additions rather than matched by guesswork.
• A plasmid written from another origin is lined up first. A circle has no first base, so the same plasmid saved by two programs can share no text at all. The dialog works out how the other file's copy has to be turned — rotated to another origin, read from the other strand, or both — says so above the differences, and compares against it turned. What is left is the real difference between the two molecules, not the difference between two ways of writing one.
• The two checksums are at the top. Each is the molecule's SEGUID v2 name, which does not change with the origin or the strand, so two that match mean the same molecule for certain — anything the dialog then lists is annotation, not sequence.
• Two unrelated sequences come out as "too different to follow in detail" rather than as a thousand tiny differences. Where nothing long enough is shared, nothing can be lined up either, and the dialog says that rather than guessing.
• Mark in the views closes the dialog and marks the same differences in the sequence view and on the map, as the Edits menu marks your own changes: the other side, lined up as the dialog lined it up, becomes what this document's marks are measured from. The Edits menu then offers Compared with _name_, and Next change (Alt+N) and Previous change (Alt+Shift+N) walk from one difference to the next. It belongs to this document: another tab keeps its own marks, and a reload brings Since opened back.

Sharing a link

File ▸ Copy share link (Alt+L) puts a link to the document on your clipboard. Paste it into an email or a chat, and whoever opens it gets the document — sequence, topology, features and all — in their own copy of PlasmidPop. The page says "Opening the shared sequence…" for the moment it takes to unpack the link, which a slower phone can take a second or so over.

Nothing is uploaded. The document travels inside the link itself, in the part after the #, which browsers never send to a server: not to the site the app is served from, not to us, not into anyone's logs. There is no account, no expiry and no server holding your plasmid, because there is no server.

What follows from that is worth knowing.

• Anyone with the link can open it. The link _is_ the document, not a pointer to it. Treat it the way you would treat the file.
• A link cannot be withdrawn or updated. It is a snapshot of the document as it was when you copied it; edit the document and copy a new one.
• Links are long, and chat apps do not all carry them. An ordinary annotated plasmid makes one of four to twelve thousand characters. Slack refuses to send a link that long, and Teams sends it but covers the screen with its tooltip when the pointer passes over it, so it is hard to click. Past 2,000 characters the notice says so. For those apps, download the file and attach it, or share a link to just the part that is needed (below). In plain-text mail such a link wraps badly — paste it as a link where you can.
• File ▸ Copy link to selection makes a shorter one: a link to just the selected bases, as a linear document of their own with the features that fall inside them — what Export selection as GenBank writes. It may wrap the origin of a circular sequence, and a selection that reaches an end of a sticky fragment keeps that end's overhang. The notice says it is a link to the selection.
• References and comments go when they are what makes it too long. A link can carry at most 32,000 characters. When a document is over that with its REFERENCE and COMMENT blocks — the papers and notes a record from NCBI comes with — and under it without them, the link is made without them, and the notice says so and how long it would have been: _Too long with its references and comments (40,210 characters); share link copied without them_. The bases, features, topology, sticky ends, host and where it came from are always in the link.
• A document too big for a link even so is refused, with its length in the message; copy a link to a part of it, or download the GenBank file and send that instead. It is the annotation rather than the bases that fills a link, so a lightly annotated long sequence may share while a heavily annotated short one does not.
• What the reader gets is their own copy. It opens in a tab of its own with no file name, and it is marked as changed since its last download, because in their browser it has never been downloaded. There is no working copy to fork, since there is no file behind it, and nothing they do can reach back to you.
• The link leaves the address bar as soon as the document opens, so the sequence is not left in the browser's history. Reloading is safe: the document is in that browser's storage like any other.

Exporting

From the File menu:

• Export figure…: the circular map or the sequence rows as a picture for figures, slides and posters. One dialog, with a preview of what will be written and its size under it, asks three things: - What — Map or Sequence view; a protein, which has no map, gets the sequence view alone. For the map, what is on it: the restriction enzymes (the cut sites of the enzymes ticked in the Enzymes tab — none while the toolbar's Cut sites toggle is off), the GC ring, the name and length in the middle, and which feature types are drawn. These options belong to the figure only; the map on screen is not changed. For the sequence view, which bases: Whole sequence, Selection (chosen for you when there is one, and highlighted in the file) or From–to, two base numbers counted from 1 as the ruler counts, both included. On a circular sequence a "to" before "from" runs through the origin — 4301 to 60 of a 4,361 bp plasmid is its last 61 bases and then its first 60, drawn as two stretches one under the other, each numbered as in the document. So does a selection across the origin. The rows are whole rows, so the first and last may hold a few bases either side of the range. The rows — ruler, bases, features and cut sites, and a read's trace — are laid out as on screen: the figure follows the Complement, Translations and Cut sites toggles, the Edits marks, the enzymes ticked in the Enzymes tab and the Format menu's complement numbering and base colours; the text size is always the export's own. The preview shows the first four rows. - How big — for the map, the text size; for the sequence view, Bases per row, 10 to 200 (the Format menu's setting to start with, or 60 when Format is left at "Fit", so the file does not depend on the window width). Then the file type: SVG (the default), a vector file that is sharp at any size; PNG, a picture at a scale of 1×, 2× or 4× the SVG's pixel size; or PDF, a document at the figure's own size (A4 or US Letter when split into pages) that is either Vector — text and lines as in the SVG, set in the PDF's standard Courier and Helvetica so the text can be searched and selected — or Raster, each page a picture at the chosen scale, as the PNG would be. The size in pixels is shown under the preview. The background is white or transparent for every type. - Pages — Split into pages, for the sequence view, on the paper chosen under Paper size: A4 (210 × 297 mm) or US Letter (215.9 × 279.4 mm, 8.5 × 11 in), with 15 mm margins. The paper size starts as US Letter when the browser's language is from the United States, Canada, Mexico or another Letter country and as A4 otherwise; nothing is remembered, and it is greyed out until pages are asked for. Pages hold as many whole rows to a page as fit, scaled down to the page's width when the rows are wider and never scaled up. Each page's foot names the document, the bases on it and the page number. The dialog says how many pages it will be before you export, and they leave as one download: a PDF holds them as its pages; SVG and PNG pages come as one ZIP file holding …_p01.svg, …_p02.svg and so on (.png for a PNG). Even the longest sequence the export takes (100,000 bases at 10 to a row, about 830 A4 pages) goes in one export as SVG, PNG or a vector PDF. Pictures are limited by their pixels, summed over the pages: a ZIP of PNGs holds up to 3,000 megapixels (all 830 pages at 2×), and so does a raster PDF. A raster PDF is slow because every page is a picture the PDF writer must re-encode, about a fifth of a second per megapixel — the 830 pages at 2× take some ten minutes, 139 A4 pages at 2× about two; the pages hint says so ("about 2 minutes to make") whenever it is a minute or more. The vector PDF of the same pages takes seconds.

While the file is made, the dialog says so beside the buttons, with a spinner and the page it is on: "Getting your file ready: drawing page 12 of 139…", then "writing the file…" (a raster PDF counts its pages a second time as they are written). The pages are drawn between the browser's turns and the file itself — the ZIP, or the raster PDF — is put together in the background, so the rest of PlasmidPop keeps answering while a long export runs. Cancel (or Escape) stops an export under way: nothing is downloaded and the dialog stays as it was, so you can change a setting and export again; press it again to close.

The dialog checks as you change things. A range outside the sequence, a row width outside 10–200, too many pages, too many pixels in the PNG or raster PDF pages, or a PNG or raster PDF larger than a browser can draw (16,384 pixels on a side) is said in red beside the Export button, which stays off until it is put right; the message says what would fit — a smaller scale, Vector, more bases per row or fewer pages. Sequences longer than 100,000 bases are refused: export a range instead. The PDF writer is fetched the first time a PDF is exported and kept for offline use after that; if you are offline that first time, the dialog says so.

• Export sequence as FASTA.
• Export read as FASTQ, for a document opened from an AB1 or FASTQ file: the bases with their qualities. Offered only while the bases are the read's; see Downloading a read.
• Export selection as GenBank or as FASTA: just the selected bases, with the features that fall inside them trimmed to the selection. The selection may wrap the origin of a circular sequence; a site feature (a point between two bases) at the origin is kept when the selection runs across it, and left out at the selection's own ends. A selection that reaches an end of a linear fragment keeps that end's sticky overhang.

The Translate tab has its own Export FASTA for six-frame translations, see Translation.

Local storage and recent files

Every open document is written to the browser's storage (IndexedDB) half a second after each change, and again when you leave the page, so a closed tab or a crash loses nothing. Its undo history is written with it, so undo and redo still work after a reload. On the start screen (the Files tab), Recent files lists these documents with size, topology, feature count and when they were last changed. Opening the same file twice does not create a second entry: a file identical to a stored document whose tab is closed opens as that document, with its history, as if you had reopened it from the list. (If you edit the new tab back to the stored contents before it is first saved, it keeps an entry of its own instead, so neither history is lost.)

• Click an entry to reopen it in a tab, or to go to its tab when it is already open. A working copy comes back as a working copy, under the name it was given, and the file it came from is still untouched.
• Rename changes the stored name (and the document name, as an undoable step of its history, whether it is open or not).
• Remove deletes it from the browser, with its undo history, and closes its tab. This does not touch files on disk.
• File ▸ Show files goes back to this list while keeping every tab open.

The view switcher and the Complement, Translations and Cut sites toggles are remembered in the same browser storage, see Viewing and selecting. So is your primer list, My primers, which leaves the browser only as a CSV or FASTA file you download.

Local storage is per browser profile and per device. A browser may clear a site's storage when disk space runs low, so the first time a document is written there a banner under the toolbar offers to ask the browser to keep PlasmidPop's storage instead. Nothing about that changes where your data is: it stays in this browser and nothing is uploaded.

• Keep my documents puts the question to the browser. In Firefox that is a dialog of the browser's own, worded something like "Allow this site to store data in persistent storage"; saying yes there means your documents are not among the first things thrown away. Chrome and Edge show no dialog and decide for themselves: they agree when the app is installed, bookmarked or used often, and otherwise say no in silence. If the browser says no, the banner tells you so, and asks again quietly on later visits until it agrees, so installing the app later is enough.
• Not now leaves things as they are. The browser is not asked, now or on later visits, and storage stays clearable.

Either way your documents are still in the browser. Clearing site data removes them, and a private window keeps nothing. So: this browser is where your work lives while you are working, and on its own it is not a backup. Download the documents you want to keep as files, or back everything up at once, below.

Backing up and restoring everything

The browser profile can go away: site data cleared, a new computer, a reinstalled system, or Safari dropping the storage of a site you have not visited for seven days. File ▸ Back up all documents… (also a button on the Files tab, under Storage and backup) writes one ZIP file, plasmidpop-backup-YYYY-MM-DD.zip, with every stored document and its undo history, and the rest of what this browser keeps for you: your primers, your parts, your own enzymes and imported tables, and the cloning shelf. It is an ordinary download, so keep it somewhere other than this browser. The ZIP is plain JSON inside; nothing is uploaded.

File ▸ Restore from backup… reads such a file back, here or on another computer. It merges and never deletes:

• A document that is not here is added, with its undo history.
• A document that is here is replaced only by a _newer_ copy from the backup. If the copy here is the same or newer it is kept, and the report says how many. So restoring an old backup never takes back later edits.
• A document that is open in a tab is left alone; close its tab and restore again to bring the backup's copy.
• Primers, parts and the other collections are added where missing; a row that is already here is kept, except the shelf and imported tables, where the newer one wins.
• A backup made by a newer version of PlasmidPop is refused with a message; a damaged entry is skipped and listed, and the rest is restored.

A backup is not opened by Open file…: that opens a ZIP of sequence files as new documents (Opening several files, or a ZIP).

Storage and backup, on the Files tab, says how many documents are stored, whether the browser has agreed to keep them, and when you last backed up. It adds a warning when there are changes older than seven days that no backup holds. In Safari, and any browser on an iPhone or iPad, that is not installed to the home screen (or the Dock), it also says that the browser may drop the documents after seven days without a visit, which a backup or installing the app protects against. The date of the last backup is remembered in this browser only, so a restore on a new computer starts with "never backed up" until you back up there.

Usage statistics

The public build sends anonymous usage statistics to a self-hosted Matomo instance: page views and coarse events such as "opened a GenBank file", "ran a ligation" or "exported the map as SVG", and which parts of the app a visit used at all — which sidebar tabs, views, toggles and guide pages, which kinds of edit, which keyboard shortcuts — each reported once per visit however often it is used. The version of the app, and whether it runs on a phone or as an installed app, are sent once. Never sent: sequences, feature names, file names, names you give your own enzymes, hosts or standards (they are reported as "custom"), sequence lengths, where in a sequence you worked, or anything else from your documents. The page address is reported without the part after the #, so opening a share link sends the tracker the app's address and nothing of the document it carries. A record opened from NCBI is counted as one (or, if it failed, why: not found, offline and so on), never by its accession. The tracker sets no cookies and the instance anonymises IP addresses. Nothing is sent at all if your browser sends a Do-Not-Track or Global Privacy Control signal, or if you switch statistics off: the Send anonymous usage statistics checkbox at the foot of this guide (Help) turns them off at once and back on without a reload, and your choice is remembered in this browser. When the browser's signal is on, the checkbox shows off and cannot be turned on. Builds without a configured instance never send anything and show no checkbox.

If Chrome asks whether the page may access other devices on your local network, that is this tracker: the statistics server is on the lab's network, and Chrome asks before a public page may reach a private address. Refusing costs nothing but the statistics; the editor itself uses the network only for Open from NCBI…, when you ask it to.

Offline use

PlasmidPop is a progressive web app. After the first visit it works without a network connection, and the browser offers to install it. When installed, sequence files can be opened with it from the file manager.

Opening an attachment from another app

On Android, with PlasmidPop installed from Chrome (the menu's Install app or Add to Home screen), PlasmidPop appears in the Share sheet of other apps. To open a GenBank file someone mailed you, share the attachment from the mail app (or a file from the Files app, or a chat) and pick PlasmidPop: it opens with the file in a tab of its own, as if you had picked it with File ▸ Open. Every format in the table above works, and several files shared at once open in a tab each. Mail apps often label a .gb file as plain text or as unknown data, so PlasmidPop is offered for those too and decides from the file itself; a file it cannot read is reported in the status bar as when opening one.

The file goes from the other app to PlasmidPop on the phone and nowhere else: it is handed over inside the browser, kept only until it has been opened, and never uploaded. iPhone and iPad do not support this (Safari has no share target for web apps); save the attachment to Files and open it with File ▸ Open instead. Desktop browsers do not offer it either.